Landslide-risk scenario of the Costa Viola mountain ridge (Calabria ...

1 downloads 0 Views 2MB Size Report
Oct 31, 2015 - Landslide-risk scenario of the Costa Viola mountain ridge (Calabria,. 1. Southern Italy). 2. Oreste G. Terranova1, Stefano L. Gariano2,3, ...
1

Landslide-risk scenario of the Costa Viola mountain ridge (Calabria,

2

Southern Italy)

3

Oreste G. Terranova1, Stefano L. Gariano2,3, Claudia Bruno1, Roberto

4

Greco1, Annamaria D. Pellegrino4, Giulio G.R. Iovine1*

5

1) CNR-IRPI (National Research Council, Research Institute for Geo-hydrological

6

Protection), via Cavour, 6, 87036, Rende (Cosenza), Italy.

7

2) CNR-IRPI (National Research Council, Research Institute for Geo-hydrological

8

Protection), via della Madonna Alta, 126, 06128, Perugia, Italy.

9

3) University of Perugia, Department of Physics and Geology, via A. Pascoli, 06123,

10

Perugia, Italy.

11

4) Regional Basin Authority of Calabria, Cittadella Regionale, Germaneto, 88100,

12

Catanzaro, Italy.

13

* corresponding author, tel. +39 0984 841 419, e-mail [email protected]

14 15

16

Landslide-risk scenario of the Costa Viola mountain ridge (Calabria,

17

Southern Italy)

18

The study area of the Costa Viola mountain ridge is strongly exposed to shallow

19

landslides triggered by rainfall. Starting from a susceptibility map, recently

20

published by the same team, the related risk has been assessed, limitedly to

21

landslide sources. The hazard has first been evaluated by considering the

22

recurrence periods of triggering events. Economic values and physical

23

vulnerabilities of the types of elements at risk have been assumed by considering

24

land-use, and effects of similar events in Southern Italy, respectively. The

25

1:30,000-scale risk map has been obtained by combining the above-mentioned

26

layers. Low risk prevails in the study area, whilst highest values are concentrated

27

along the coast, where villages and main infrastructures are located. The expected

28

annual damage exceeds 570 Million € (about 68% pertaining to low-risk cells).

29

Risk maps obtained through similar simplified GIS-based procedures may play a

30

crucial role in accurate land-use planning and effective decision-making.

31

Keywords: shallow landslide, susceptibility, hazard, physical vulnerability,

32

potential worth of loss, risk, logistic regression

33 34

1. Introduction Shallow landslides are among the most damaging natural hazards in terms of

35

economic losses and casualties. Such phenomena are characterized by rapid dynamics

36

and are widespread worldwide. As for other types of dangerous phenomena, in addition

37

to direct damage – e.g., those caused to buildings, transport infrastructures, agricultural

38

crops, water supply networks, and sewerage and communication systems – indirect

39

damage should be considered: for instance, conspicuous financial losses may occur due

40

to impossibility of using assets not directly affected by the landslides. Events

41

characterized by activation of widespread shallow landslides may hamper the economic

42

and social development for years after their occurrence. Furthermore, the same area

43

may be affected by recurrent landslide events due to its intrinsic exposition to

44

predisposing (e.g., steep slopes, lithology) and triggering factors (e.g., intense rainfall,

45

seismic shocks).

46

Aiming at assessing hazard and risk, the evaluation and mapping of spatial triggering

47

probability of shallow landslides – i.e., “source susceptibility” – must first be

48

accomplished. In case a significant propagation downslope of the landslide source is

49

expected (e.g., in case of soil slip-debris flow transition), the “propagation

50

susceptibility” must also be evaluated (Iovine, 2012). The occurrence of shallow

51

landslides is usually strictly related to external triggering factors: therefore, hazard

52

evaluations can be performed based on recurrence probabilities of the causal factor

53

(e.g., rains or earthquakes), which can be estimated by statistical analyses of temporal

54

series. Finally, when vulnerability and economic values of the elements exposed to risk

55

are known, a further, multidisciplinary step toward total risk assessment can be

56

undertaken (Varnes and IAEG Commission on Landslides and Other Mass Movements

57

on Slopes, 1984; Dai et al., 2002; Gaprindashvili and van Westen, 2016).

58

Due to its geological history, and to present climatic and seismic characteristics,

59

Calabria (Southern Italy) is mostly prone to shallow landslides (Carrara et al., 1982;

60

Pellegrino and Borrelli, 2007). In fact, prolonged rainfall and rainstorms frequently

61

occur; earthquake catalogues include several high-magnitude events occurred in the past

62

centuries; weathered and tectonized terrains diffusely crop out along steep slopes cut by

63

the hydrographic network on actively uplifted massifs. In Calabria, shallow landslides

64

are mostly triggered by severe rains (Terranova and Gariano, 2014; Vennari et al.,

65

2014). Frequency and intensity of rainstorms increased in recent years, despite a

66

reduction of annual rainfall amounts, due to occurrence of medicanes favoured by the

67

interaction between warm sea and rugged orography. A recent investigation allowed to

68

identify numerous landslide events that affected either small areas or large part of the

69

Calabrian territory in the last decades (cf. Iovine et al., 2014, and references therein).

70

In this study, an example of rainfall-induced shallow-landslide risk evaluation at source

71

locations is proposed for the Costa Viola mountain ridge (Calabria). The analysis is

72

based on “source susceptibility” evaluations, given the limited runout generally

73

observed in historical cases occurred in the study area. Recurrence periods of the

74

triggering events were obtained through hydrological analyses of annual maxima of

75

daily rainfall. Economic values of the elements at risk were assumed based on land-use.

76

Physical vulnerability was obtained by taking into account types of existing elements

77

and intensities of expected phenomena, together with effects of similar events occurred

78

in Southern Italy. The resulting risk map (at 1:30,000 scale) has been obtained by

79

properly combining, in a GIS environment, the above-mentioned layers.

80

2. Study area

81

The Costa Viola mountain ridge extends approximately 82 km2 along the SW

82

coast of Calabria. The area includes short high-gradient drainage basins (the widest

83

thereof are named Favazzina and Sfalassà), which flow into the Tyrrhenian Sea between

84

the villages of Scilla and Bagnara Calabra (Figure 1). The Palaeozoic basement of the

85

area is mainly made of middle-high grade metamorphic rocks and acid intrusive rocks,

86

strongly tectonized and deeply weathered (Figure 2). The sedimentary cover (Late

87

Miocene to Holocene deposits) unconformably lies on the basement, and is made of:

88

conglomerate, arenite, clay, and marl with subordinate evaporite (Late Miocene-Middle

89

Pliocene); sand and conglomerate, generally terraced (Middle Pliocene-Middle

90

Pleistocene); alluvial and coastal gravel, sand, silt and clay (Late Pleistocene-

91

Holocene).

92

At the base of the mountain ridge, a NE-SW trending fault system is to be found. Such

93

tectonic structures belong to the Calabrian-Sicilian Rift Zone, an active regional fault

94

system that permitted the uplift of the Costa Viola with respect to the Tyrrhenian basin

95

since Middle-Pleistocene (Tortorici et al., 1995). In the area, seven orders of marine

96

terraces were recognised along the Tyrrhenian coast (Tortorici et al., 2002 – cf. Figure

97

2b), as a result of the combined effects of regional uplift and eustatic sea-level changes.

98

Extensive flat areas characterize the study area, bordered by steep slopes and cut by

99

deep canyons. Short high-gradient torrents, characterised by ephemeral discharges and

100

flash floods, drain the western slope of the ridge. Along the densely urbanized coast,

101

steep cliffs delimit outcrops of narrow and discontinuous deposits (alluvial/colluvial and

102

marine).

103

In the area, average yearly rainfall varies between 600 and 900 mm·y−1 (Terranova,

104

2004; Terranova and Iaquinta, 2011). At the Scilla rain gauge, located by the coast and

105

characterized by the longest available data record (from 1939 to 2015), annual amounts

106

range from 507.2 mm to 1270.6 mm (782.3 mm on average), mainly concentrated

107

between October and March (ca. 74% of the annual amount). The maximum daily

108

rainfall (147.8 mm) was recorded at Scilla-Solano on November 1, 2015.

109

The study area is scarcely urbanized. The only villages are Bagnara Calabra and Scilla:

110

the first is located in the widest portion of the above-cited coastal deposits; the ancient

111

settlement of Scilla lies on a small rocky outcrop facing the sea, whilst its recent

112

hamlets mainly developed by the coast. The main infrastructures are mostly to be found

113

in the coastal sector: here, the railway, the highway (A3 “Salerno-Reggio di Calabria”),

114

and the southern Tyrrhenian state road (SS.18) extend in a narrow strip.

115

In the study area, shallow landslides are mainly represented by debris slides and earth

116

slides (often called “soil slips”, sensu Campbell, 1975; see also Cruden and Varnes,

117

1996), that commonly show a limited run-out (ten metres in average). Subordinately,

118

rock falls and debris flows are also to be found, characterized by greater run-out and

119

destructive power; larger slope movements of various types are also present. As a

120

whole, the average run-out of the mapped landslides is ca. 50 m.

121

In the last 75 years, several slope instability events affected the study area (on average,

122

ca. 0.6 events per year, cf. Figure 3), mainly related to debris slides, rock falls and,

123

subordinately, debris flows, that damaged rural areas and the secondary road network.

124

In the last 20 years, a remarkable increase of frequency of damaging events can be

125

appreciated (up to ca. 1.3 events per year). On 12 May 2001 and 31 March 2005, in

126

addition to numerous debris slides that stroke the innermost portion of the study area,

127

debris flows severely hit the urbanized areas and the main transportation infrastructures,

128

causing train derailment (Bonavina et al., 2005). As a rough underestimation of the

129

economic loss produced by shallow landslides (except for debris flows) since 2001, the

130

funds for remedial works allocated by either national or regional authorities amount to

131

ca. 9.5 Million € (Basin Authority of Calabria – unpublished data).

132

Despite evidence of hydraulic erosion and shallow instabilities on the slopes, the return

133

periods (T) of the rainfall recorded at the Bagnara Calabra and Scilla rain gauges in the

134

last two decades are rarely notable, presumably due to a different (“unlucky”) location

135

of the rain gauges with respect to the rain cell centres. For instance, those related to the

136

events occurred in 2001 and 2005 are not worthy of note (T < 2y for daily maxima);

137

low values (T~3y) also characterize rainfall recorded from 2008 to 2010, with T Tr are listed for

210

Scilla and Bagnara Calabra: a marked spatial variability of rainfall can be appreciated

211

on some cases; values of T range from 10 to over 100 y. More in detail, the historical

212

maximum occurred in 1988 (i.e., 2-3 years before the flight used for validation), whilst

213

maximum daily rainfall recorded in 1948 e 1951 have T = 20.4 and 24.4 y, respectively.

214

Annual maxima of daily rainfall from 1947 to 1955, and from 1974 to 1988, may be

215

considered responsible for the slope movements recognized in the 1955 and the 1990

216

flights (i.e., those employed for calibration and validation), respectively. As a result, the

217

susceptibility map obtained by taking into account the calibration set of shallow

218

landslides can be related to triggering rains with about 20 y of return period, therefore

219

depicting a hazard scenario (H) with T = 20 y (cf. H in Figure 5).

220

The hazard was derived from the original (unclassified) susceptibility map. In

221

theoretical terms, the probability of occurrence, P, of an event (e.g. a rainfall able to

222

trigger a given type of landslide) can be obtained from its return period, T, as: P  1 / T .

223

Hazard related to an event of given intensity can also be computed from the classical

224

formula H  1  1  1

225

(cf. e.g. Gostelow, 1992). Nevertheless, information needed for such type of evaluations

226

is commonly inadequate in landslide analyses at regional scale. Hence, either

227

approximated or empirical approaches must be adopted. In the present study, hazard

228

was merely obtained from susceptibility values, by considering the ca. 20 y return

229

period of the triggering rainfalls responsible for the landslide population used for

230

calibration: susceptibility values were, then, adopted as a proxy for hazard with respect

231

to a scenario of 20 y, being assumed as representative for the considered temporal

232

window.

233

Economic values (W) of the elements at risk were assumed (Table 5, second column)

234

depending on land use. Values were assigned based on market surveys (mainly for



 P

N

, in which N is the extent of the considered temporal window

235

buildings), construction costs for transport infrastructures, and asset values compared

236

with results of market surveys for agricultural and urban uses. Note that the adopted

237

economic values span in a range of four orders of magnitude.

238

By taking into account previous events occurred in the same environment (Bonavina et

239

al., 2005; Iovine et al., 2014), the potential propagation of the soil slips in form of debris

240

flows (run out) could be neglected. In addition, based on intensities of expected

241

phenomena and on effects of previous shallow-landslide events, occurred in Southern

242

Italy in the past decades (Sarno, May 1998; Soverato, September, 2000; Messina,

243

October 2010 – cf. Antronico et al., 2002; Sorriso-Valvo et al., 2004; Iovine et al.,

244

2009; Ardizzone et al., 2012), physical vulnerabilities (V) were assumed for the types of

245

elements at risk recognized in the study area (e.g., buildings, roads, agricultural crops –

246

cf. Table 5, third column). Vulnerability values, determined by expert knowledge, range

247

from 0.1 to 0.9.

248

Accordingly, the potential worth of loss ( WL  V  W ) could be computed (Table 5, last

249

column; cf. WL in Figure 5). By considering the variability in space and time of land use

250

in the study area, values of WL were ranked into 4 classes: WL< 1; 1≤WL