Geomorphology 126 (2011) 148–158

Contents lists available at ScienceDirect

Geomorphology
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / g e o m o r p h

Spatial and temporal occurrence of rainfall-induced shallow landslides of flow type:

A case of Sarno-Quindici, Italy
Leonardo Cascini, Sabatino Cuomo ⁎, Maria Della Sala
University of Salerno, Department of Civil Engineering, Via Ponte Don Melillo, 1 – 84084 Fisciano (SA), Italy

a r t i c l e i n f o a b s t r a c t

Article history: Rainfall-induced shallow landslides of flow type provide unstable masses which often travel long run-out
Received 26 December 2009 distances with high velocities, thus posing a high societal risk when they affect large areas. Therefore, analysis
Received in revised form 22 October 2010 of their spatial and temporal occurrence is relevant to landslide hazard assessment as the first step of the risk
Accepted 31 October 2010
analysis. In order to address this issue, this paper outlines a multidisciplinary procedure that is applied to the
Available online 6 November 2010
May 1998 Sarno-Quindici landslides (southern Italy), whose spatial and temporal occurrence is not
satisfactorily addressed in current literature. The spatial occurrence of the landslides is analysed using
Keywords:
Shallow landslides
heuristic models for both the source and propagation areas. The temporal occurrence of the landslides is firstly
Flow type compiled and then related to the cumulated rainfall, stratigraphy and hydraulic boundary conditions. Finally,
Rainfall the spatial and temporal occurrence of the main landslide triggering mechanism is modelled over the whole
Spatial and temporal occurrence affected area by analysing the groundwater regime and slope stability conditions. The obtained results show
Modelling that the spatial and temporal occurrence is strongly related to stratigraphy and hydraulic boundary conditions
at both the slope and massif scales. They also highlight a suitable procedure for assessing the spatial and
temporal occurrence of complex landslides over large areas.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction single slope sections and more detailed stratigraphies. Landslide

Rainfall-induced shallow landslides of flow type (Hungr et al.,
2001) can travel long run-out distances (hundreds of meters), often
with high velocities on the order of m s− 1. When they occur over a
large area, they may pose a high societal risk (Cascini et al., 2008a) as
testified by several examples all over the world (Fuchu et al., 1999;
Wang et al., 2002; Capra et al., 2003; Lacerda, 2004; Take et al., 2004).
Understanding these destructive phenomena through the analysis of
the relationships among their spatial/temporal occurrences and the
predisposing/triggering factors is important in order to reduce their
negative consequences.
In literature, the distribution of landslide source areas is analysed
using different models commonly implemented on GIS platforms (van
Westen, 2004). The so-called geological models relate the distribution
of landslide source areas to combinations of spatial variables such as
slope angle and thickness of deposits using either heuristic (Dai and
Lee, 2002) or statistical methods (Dai and Lee, 2002; Corominas et al.,
2003). More advanced models simulate the hydrological response of
hillslopes and evaluate their stability conditions. Among these,
physically based models (Dietrich and Montgomery, 1998; Godt et
al., 2008) analyse large areas using the simplified scheme of infinite
slope, while geomechanical models (Cascini et al., 2010) refer to
propagation areas are mainly analysed through mathematical models
which simulate landslide propagation stages (Pastor et al., 2009) and
empirical models which relate the run-out distance to topography and
the volume of unstable mass (Corominas, 1996). In contrast, the
temporal occurrence of shallow landslides is addressed independent-
ly. Either black-box models are used which relate the failure onset to
the amount of cumulative rainfall (Caine, 1980) or hydrological
models which simulate the effects of rainfall through analytical
relationships (Sirangelo and Braca, 2004).
The correlation between the spatial and temporal occurrences of
shallow landslides over large areas can be analysed using physically
based models. However, the success and error of these models strictly
depend on both the characteristics of the landslide source areas and
triggering mechanisms (Sorbino et al., 2009). As a consequence, a
general approach is not yet available in current scientific literature to
tackle this issue. This paper contributes to the topic investigating a
relevant case (May 1998 Sarno-Quindici event, southern Italy) for
which an advanced geotechnical dataset is available.
2. Studied event and methods
2.1. The event

The May 1998 event (Cascini, 2004) is among the most

⁎ Corresponding author. Tel./fax: + 39 089 964231. catastrophic recent natural disaster in the Campania region which is
E-mail address: scuomo@unisa.it (S. Cuomo). one of the most landslide-prone areas in Europe (Cascini et al., 2008a).

0169-555X/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.geomorph.2010.10.038
 L. Cascini et al. / Geomorphology 126 (2011) 148–158 149

Fig. 1. The Sarno-Quindici study area affected by the May 1998 flow-type landslides.

This event was characterised by tens of flow-type landslides over a
60 km2 massif (Cascini et al., 2005, 2008b; Guadagno et al., 2005)
(Fig. 1). The affected area consists of shallow (depth b 5) unsaturated
pyroclastic air-fall deposits, originating from the explosive activity of
the Somma-Vesuvius volcanic apparata (Cascini et al., 2008a). The
bedrock below is a fractured and karsified carbonate massif,
characterised by a suspended groundwater flow system in its upper
part and springs at different altitudes along the hillslopes (Cascini et
al., 2008b). The hillslopes contain three typical geomorphological
units: zero order basins (zobs), open slopes and flanks of valley
(Cascini et al., 2008b). Zero order basins are colluvial hollows with a
concave bedrock profile characterised by a maximum depth in the

Fig. 2. Three types of landslide source areas (M1, M2 and M3). a) schematic illustration of the three types. 1: bedrock, 2: pyroclastic deposits, 3: track, 4: spring from bedrock. b)
distribution of the three types and geomorphological units affected (1) and not affected (2) by the May 1998 landslides.
150 L. Cascini et al. / Geomorphology 126 (2011) 148–158

Table 1
Morphometry of the main geomorphological units.

Landforms Morphometry Pyroclastic deposits

Slope angle Length Width Area Depth Volume

(°) (m) (m) (m2) (m) (m3)

15th perc. Mean 85th perc. Mean Mean Min Max Min Max Min Max

Zero order basin 29 32 35 295 111 5413 69,889 2 4.5 10,826 314,500
Open slope 20 28 34 357 519 32,586 411,265 1 4.5 32,586 1,850,696
Flank of valley 30 34 39 109 603 1203 169,432 1 4.5 1203 762,444

central part. The open slopes have a nearly constant slope angle and
depth of the pyroclastic deposits. The flanks of valley are hillslopes
with a channel at the lower part.
Concerning the spatial occurrence of the May 1998 landslides,
Cascini et al. (2008b) note that failures occurred in all of the three
geomorphological units, with the corresponding landslide source areas
being referred to as M1, M2 and M3, respectively. M1 inside the zobs is
characterised by elongated planforms. M2 on open slopes has triangular
planforms. M3 on the flanks of valley show compound landforms.
The triggering mechanisms of the landslides in the M1, M2 and M3
source areas are also labelled as M1, M2 and M3, respectively. The
triggering mechanism M1 is related to rainwater infiltrating the
ground surface and temporary springs from the bedrock. The
triggering mechanism M2 is caused by springs from karst conduits
and/or impact of small landslides occurred at the top of bedrock
scarps. The triggering mechanism M3 is related to rainfall infiltration
and concentration of runoff water in particular zones close to
mountain roads or tracks. All these mechanisms were analysed by
Cascini et al. (2008b,c,d) for the geomechanical modelling of the
failure stage. It should be stressed that the different landslide
triggering mechanisms in different geomorphological units reduce
the ability of physically based models as shown by Sorbino et al.
(2009). Similar difficulties arise when geological and geomechanical
models are used.
For the propagation areas, Budetta and de Riso (2004) indicate that
the May 1998 landslides had a higher mobility than analogous slope
instability phenomena in other areas of the Campania region (i.e.
Sorrento Peninsula). The mobility of these landslides is also addressed
by Pareschi et al. (2002) through geomorphological analyses as well
as Revellino et al. (2004) and Pastor et al. (2009) through numerical
modelling. However, the relationship between the run-out distance
and triggering mechanisms has not been discussed in the current
scientific literature.
Relative to the temporal occurrence, Rossi and Chirico (1998) and
De Vita (2000) relate the onset of the May 1998 landslides to both
critical 2-day as well as antecedent cumulative rainfall values.
Sirangelo and Braca (2004) propose a hydrological model to explain
the temporal occurrence of the May 1998 landslides that are analysed
as a single event, while Cascini et al. (2003) demonstrate that multiple
failures occurred inside a single landslide source area over a time
period of about 10 h. These contributions, however, do not discuss the
temporal occurrence of these landslides in the whole affected area.
2.2. Adopted procedure
This paper jointly analyses the spatial and temporal occurrences of
landslides, considering the main geological and geotechnical aspects at
both the slope and massif scales. With this aim in mind, the adopted
procedure addresses four main questions: i) Is it possible to discriminate
between geormorphological units affected and not affected by failures,
on the basis of their features before the event? ii) Is the landslide run-out
distance related to the location and features of the source areas? iii) Is

Fig. 3. Spatial distribution of pyroclastic soils and typical stratigraphy types (1–4) of the pyroclastic air-fall deposits for each sector of the massif.
 L. Cascini et al. / Geomorphology 126 (2011) 148–158 151

Table 2
Geomorphological units affected by failure in relation to stratigraphy.

Sector of the Stratigraphy Geomorphological Geomorphological Geomorphological units (number) Geomorphological units affected by
massif (type) units (number) units affected by failure (%)
failure (%)
Zob Open slope Flank of valley Zob Open slope Flank of valley

Bracigliano-Siano 1 45 33% 29 8 8 38% 13% 38%

Quindici 2 or 3 36 44% 12 0 24 33% – 50%
Sarno 3 or 4 97 18% 38 15 44 29% 13% 9%

the temporal landslide occurrence related to the predisposing and
triggering factors? iv) Are the spatial and temporal occurrences of the
landslides independent?
In order to answer these questions, heuristic analyses are initially
carried out. In particular, the onset of landslides in their source areas is
analysed in relation to the morphometric and stratigraphic character-
istics of the main geomorphological units before failure. Subsequently,
the run-out distance is related to the location of the source areas and
triggering mechanisms. The failure time sequence is also investigated
in relation to the landslide triggering mechanisms. Finally, the spatial
and temporal occurrences of the landslides are concurrently analysed
through the geomechanical modelling of their failure stage in order to
evaluate the temporal variations of the slope stability conditions over
the massif.
These analyses are carried out using an extensive geotechnical
dataset obtained from current literature (U.O. 2.38, 1998; Sorbino and
Foresta, 2002; Crosta and Dal Negro, 2003; Picarelli et al., 2004; Bilotta
et al., 2005; Cascini et al., 2006). Details on the geological setting can
be found in Cascini et al. (2008b), while details on the geotechnical
properties of the involved pyroclastic soils are provided by Bilotta et
al. (2005), who identified three main lithotypes: pumice soils
essentially constituted by sands and gravels, coarser superficial ashy
silty sands (class B), and finer deep ashy sandy silts (class A).
3. Analysis of landslide spatial occurrence
3.1. Failure onset
The landslide spatial occurrence was investigated in relation to
the three main geomorphological units (Fig. 2). Among the 178
geomorphological units shown in Fig. 2, only 48 were affected by
slope failures during the May 1998 event, of which 26 inside zobs (out
of a total of 79; 33%), three on open slopes (out of 23; 13%) and 19 on
flanks of valley (out of 76; 25%).
Table 1 shows that the slope angle is the highest for the flanks of
valley and the smallest for open slopes. In contrast, the areal extension
is the largest for open slopes and smallest for zobs. The thickness of
the pyroclastic deposits is the largest in the zobs and the smallest on
flanks of valley, but the volume of the pyroclastic deposits (evaluated
by simply assuming a rectangular transversal section) is the largest for
open slopes and smallest for flanks of valley.
During the May 1998 event, zobs and flanks of valley were mostly
affected by the landslides. This observation can be related to the thickest
pyroclastic deposits inside the zobs and the highest slope angles on the
flanks of valley. However, the zobs and flanks of valley affected by the
failures have a wide range of slope angles, while only the steepest open
slopes experienced failures in most cases. Consequently, it is impossible

Fig. 4. Travel paths of the landslides selected for the analysis of the run-out distance, I) influenced and II) not influenced by anthropogenic structures at the toe of the massif.
152 L. Cascini et al. / Geomorphology 126 (2011) 148–158
to strictly relate the landslides spatial occurrence to the morphometry of
the geomorphological units, as also highlighted by Guadagno et al.
(2005) as well as Di Crescenzo and Santo (2005). Therefore, other aspects
must be taken into account such as the stratigraphy of the slope deposits.
The latter is thus discussed using 325 stratigraphical profiles, as well as
the porosity values n and specific gravity Gs of 111 laboratory specimens
of ashy soils (data from U.O. 2.38, 1998). The spatial distribution of ashy
(A and B) and pumice soils is shown in Fig. 3, where the main sectors of
the massif show four stratigraphic types: i) stratigraphy 1 with only ashy
B soil near the ground surface; ii) stratigraphy 2 with ashy B soil above
ashy A soil on the bedrock; iii) stratigraphy 3 with ashy B soil, a pumice
layer and ashy A soil; iv) stratigraphy 4 similar to stratigraphy 3 but with
two to four pumice layers.
We observed that: i) slope failures mainly involved ashy B and
pumice soils and, to a minor extent, ashy A soil; and ii) the highest
percentage (44%) of geomorphological units affected by the failures
occurred in the sector of Quindici (Table 2).
Further insights on this topic can be obtained by separately
analysing the different geomorphological units in relation to their
stratigraphy (Table 2), and considering the triggering factors in the
analysis. For example, regarding zobs with all the stratigraphical
settings (1–4), more pumice soil layers may correspond to decreased
failures. This could be essentially due to triggering mechanism M1, i.e.
rainfall infiltrating the ground surface and springs from the bedrock.
Fig. 5. Height of fall (H) and run-out (L) of landslides. a) all data. b–d) data divided according to b) sector of the massif, c) landslide typology and d) landslide source area.
In this case, continuous pumice soil layers could improve the slope
stability conditions. However, similar results cannot be found for
the other geomorphological units where mechanisms M2 and M3
occurred, meaning that other factors should be investigated.
3.2. Run-out distance
Landslide spatial occurrence was further analysed in relation to the
height of fall (H) and run-out distance (L). H and L were computed
between the uppermost landslide source area and the lowest deposition
area; L was measured along the landslide propagation path (Fig. 4).
The obtained results (Fig. 5a) highlight that anthropogenic struc-
tures such as paved roads and channels located at the toe of hillslopes,
significantly increased the landslide run-out. In particular, the reach
angle α (i.e. the arctangent of the ratio of H to L) is significantly lower
for the landslides influenced by the anthropogenic structures (Fig. 5a).
H and L are well interpreted by linear relationships as indicated by
previous studies (Pareschi et al., 2002; Budetta and de Riso, 2004).
Further insights into the run-out distances were obtained by sub-
dividing the dataset in terms of the position and type of the landslide
source areas. For instance, in the northwest sector of the massif (i.e.
Quindici and Bracigliano) the landslides had a similar mobility (Fig. 5b),
but different from those in the southern sectors of the massif (i.e. Sarno
and Siano), where the landslides exhibited longer travel distances and
 L. Cascini et al. / Geomorphology 126 (2011) 148–158
Fig. 6. Time sequence for the May 1998 landslides. Numbers refer to hours on 5th May 1998.
lower reach angles. Moreover, the propagation pattern of single
landslides and that of multiple landslides (i.e. landslides which joined
along the propagation path; Fig. 4) are also different, with the latter
having longer travel distances and lower reach angles (Fig. 5c).
Concerning single landslides, L and H show a good correlation for
M1 and M2, but not for M3 (Fig. 5d). In particular, the lowest reach
angles were obtained for M1 and comparable travel distances are
observed for M1 and M2. In conclusion, the run-out distance of the
landslides depends on both the location (i.e. the sector of massif) as
well as the features of the landslide source areas (M1 to M3).
4. Analysis of landslides temporal occurrence
4.1. Reconstruction of failures time sequence
The temporal occurrence of the May 1998 landslides was recon-
structed using the available eyewitnesses who provided the arrival time
of the failed masses at the toe of the massif (Fig. 6). These data highlight
that the landslides firstly affected the sectors of Quindici (from 5th May
11 a.m.) and Bracigliano (2 p.m.), then Siano and Sarno-Lavorate (4 p.m.)
and finally Sarno-Episcopio (8 p.m.). Moreover, inside the sectors of
Quindici, Bracigliano and Siano, the landslides occurred before 5th May
9 p.m. while in the other sectors the landslides did not occur until 12 p.m.
Due to the high velocities of the propagating masses that travelled
the paths within a few minutes, the arrival time at the end of the
travel paths nearly corresponds to the failure time in the landslide
source areas (Fig. 6). Therefore, it is possible to say that: i) 11 M3
source areas experienced landslides in about 11 h from 5th May 11 a.
m. (except for a landslide occurring after 10 p.m.), ii) 33 M1 areas did
over a longer time period (13 h after 11 a.m.), and iii) 11 M2 areas did
in a shorter time period (8 h after 4 p.m.).
4.2. Analysis of the failure time sequence
The failure time sequence was investigated by analysing the
amount of rainfall between 4th May and failure, since rainfall was a
153
triggering factor for all the mechanisms M1, M2 and M3 (Cascini et
al., 2008a,b,c). The analysis was based on twice the rainfall data
recorded at the S. Pietro rain gauge (Cascini et al., 2003) located at
215 m a.s.l. and 10 km away from the Pizzo d'Alvano massif.
The results show that the landslides in Fig. 6 were triggered
by rainfall values of 156–210 mm for M3, 156–248 mm for M1 and
192–248 mm for M2.
The failure time sequence described in Section 4.1 indicates that a
strict correspondence between the rainfall amount and the failure
onset does not exist. Consequently, other triggering factors must be
considered in order to explain the sequence. In particular, the earliest
failure occurrence and low rainfall values for M3 could be explained
by the topography that allows the rapid concentration of rain water
at the bends of the tracks. Conversely, the higher rainfall amount
required for triggering failures in M2 areas could be explained by
water infiltration into karst conduits.
More insights are provided for M1areas (Fig. 7). For instance, the
earlier landslide occurrence in the Quindici sector rather than in Sarno
could be related to the differences in the stratigraphy of the
pyroclastic deposits as well as the bedding of the bedrock layers
which affects the springs from the bedrock, earlier acting in Quindici
(Cascini et al., 2008b). Stratigraphy could also explain the later
occurrence of failures in Sarno-Episcopio compared to Sarno-
Lavorate. These inferences should be confirmed through the geome-
chanical modelling of the failures.
5. Geomechanical modelling
Following the procedure described in Section 2.2, the failure
stage of the triggering mechanism M1 was analysed by using
the simplest approach of Cascini et al. (2010). This approach is
based on the following steps: i) computation of pore water
pressures with the aid of a transient uncoupled analysis of the
groundwater regime, and ii) slope stability analyses using limit
equilibrium methods.
154 L. Cascini et al. / Geomorphology 126 (2011) 148–158

Fig. 7. Failure time sequence for the landslide source areas M1: numbers refer to hours on 5th May 1998 and circles refer to the time sequence of the events.

5.1. Slope schemes and soil properties unsaturated conditions were also discussed by Bilotta et al. (2005,
2008). The properties used are summarised in Fig. 9 and Table 3. A full
The slope schemes for the modelling were obtained from the typical parametric analysis was carried out by changing the soil shear
stratigraphical settings outlined in Section 3.1. For each sector of the strength parameters.
massif, typical stratigraphies and slope angles were considered while
different depths of the pyroclastic deposits were assumed (Fig. 8). 5.2. Groundwater modelling
The physical and mechanical properties were taken from current
literature. The volumetric water content and hydraulic conductivity We initially simulated pore water pressure changes between 1st
curves for ashy A and B soils, obtained through the suction controlled January and 5th May, 1998. A saturated–unsaturated transient
oedometer, volumetric pressure plate extractor and Richards pressure groundwater model was adopted from Richards (1931):
plate, were discussed by Sorbino and Foresta (2002), while empirical
relationships are available for pumice soils (Sorbino and Foresta,    
2002; Bilotta et al., 2005). On the other hand, the shear strength and ∂ ∂h ∂ ∂h ∂h
kx + ky = mw γw ð1Þ
compressibility properties of ashy soils in both saturated and ∂x ∂x ∂y ∂y ∂t

Fig. 8. Slope schemes for geomechanical modelling.

 L. Cascini et al. / Geomorphology 126 (2011) 148–158 155

Fig. 9. Geotechnical properties of pyroclastic soils. a) grain size distribution. b) saturated shear strength envelopes. c–d) peak shear strength of ashy A and B soils versus net vertical
stress for some ranges of saturation degree (Srf) (after Bilotta et al., 2005).

Table 3 The commercial finite element code SEEP/W (Geoslope, 2005) was
Physical and mechanical properties of the pyroclastic soils used in the geomechanical applied to a finite element mesh with 3500 quadrilateral elements
modelling.
with lengths and heights respectively smaller than 1.43 and 0.18 m
γd γsat n ksat c′ φ′ φb and the soil parameter values in Table 3. As a boundary condition at
(kN m− 3) (kN m− 3) (–) (m s− 1) (kPa) (°) (°) the ground surface, a flux condition was assumed equal to daily
Ashy soils (class A) 9.1 15.7 0.66 10− 6 5–15 32–35 20 rainfall intensity recorded at the toe of the hillslopes between the 1st
Pumice 6.2 13.1 0.69 10− 4 0 37 20 of January and 3rd of May 1998. For the last 2 days (4th–5th May), the
Ashy soils (class B) 7.3 13.1 0.58 10− 5 0–5 36–41 20
hourly rainfall intensities mentioned in Section 4.2 were assumed. The
evapo-transpiration phenomena were neglected because Sorbino
(2005) demonstrates that such phenomena do not significantly affect
where mw is the coefficient of volumetric water changes with respect the suction regime during Spring. As a boundary condition at the
to the change in negative pore pressure, and it is equal to the slope of bedrock contact, an impervious condition is assumed over the slope
the soil–water characteristic curve; h is the total head; kx and ky are section except where a spring from the bedrock has been operating
the soil conductivity coefficients in x and y directions; γw is the unit with a flux of 8.57 × 10− 3 m3 s− 1 since 3rd of May 1998 (Cascini et al.,
weight of water; and t is time. 2003). As an initial condition, a uniform distribution of suction equal

Fig. 10. Typical groundwater regime for different slope schemes after 12 h.
156 L. Cascini et al. / Geomorphology 126 (2011) 148–158
Fig. 11. Trasmissivity of different slope sections used for modelling.
to 10 kPa is assumed all over the slope section based on Cascini et al.
(2003, 2005).
The obtained results show that the simulated pore water pressure
regime significantly differs for the assumed slope schemes (Fig. 10),
especially in the zone with the spring. The highest pore
water pressures were simulated for the schemes QU, SL and SE
(i.e. where B and A soils are present) and the lowest values
correspond to the schemes BR and SI (i.e. containing only B soils).
These differences can be profitably interpreted based on the concept
of trasmissivity T:
T = ∑ ki ⋅ bi
i with the in-situ evidence. These results can be explained by the values
where i represents the number of soil layer, ki is the saturated
hydraulic conductivity and bi is the depth of each soil layer.
Cascini et al. (2003) demonstrate that T is a simple but efficient
parameter to interpret slope responses to rainfall infiltration which
strongly affects both the spatial and temporal occurrences of slope failures.
Fig. 12. Simulated unstable volumes for the slope schemes of Fig. 8.
For the assumed slope schemes of Fig. 8, T varies from 2.97 × 10− 5
to 8.37 × 10− 5 m2s− 1, corresponding to the range assumed by Cascini
et al. (2003, 2005) except the schemes SI2, BR2 and QU2. For these
schemes, the simulated high pore water pressures are associated with
the low values of T and vice versa (Fig. 11).
5.3. Slope stability analysis
In relation to the computed pore pressures, slope stability
analyses were carried out to simulate failure conditions mainly
where the spring from the bedrock is located. These local failures can
induce multiple retrogressive landslides which can entirely mobilize
the pyroclastic deposits of zero order basins (Cascini et al., 2003,
2005).
The slope stability conditions were evaluated through the limit
equilibrium methods proposed by Janbu (1954) as well as Morgen-
stern and Price (1965) by using the commercial SLOPE/W code
(Geoslope, 2005). A rigid-perfectly plastic constitutive model was
used for the involved soils referring, in both saturated and
unsaturated conditions, to the extended Mohr–Coulomb failure
criterion proposed by Fredlund et al. (1978). A parametric analysis
was carried out for the aforementioned slope schemes using the soil
mechanical properties in Table 3.
The slope stability conditions were initially computed based on the
pore water pressures simulated for the beginning of 4th May 1998. For
the schemes SL2, SL4.5, SE2, SE4.5 and QU4.5, the factor of safety (FS)
is higher than 1 only when c ' is not null for ashy A soils. If c ' is
assumed to be 15 kPa, FS is always higher than 1 irrespective of φ′ for
the ashy B soils. For the schemes SI4.5 and BR4.5, FS higher than 1 is
simulated only when φ′ is higher than 37° for the ashy B soils. For the
remaining slope schemes SI2, BR2 and QU2, FS is lower than unity,
irrespective of the soil mechanical properties, which is inconsistent
ð2Þ
of T assumed for these schemes, which are lower than those used by
Cascini et al. (2003, 2005) to satisfactorily back-analyse the 1998
landslides.
Focusing on the realistic slope schemes (QU4.5, SL2, SL4.5, SE2,
SE4.5, SI4.5, and BR4.5), the slope stability conditions were evaluated
during the time period 4–5 May 1998, at each time step (i.e. hourly),
 L. Cascini et al. / Geomorphology 126 (2011) 148–158
Fig. 13. Simulated failure time sequence for the slope schemes of Fig. 8, assuming for ashy soils A c ' = 15 kPa and φ′ = 35°, and for ashy soils B c ' = 0 kPa and φ′ = 39.5°.
by assuming different mechanical properties for the ashy soils. Fig. 12
provides the simulated unstable volumes which correspond to the
whole depth of the pyroclastic deposits for the sector of Bracigliano
(scheme BR4.5). For the schemes involving ashy A soils (schemes QU,
SE, and SL), the slip surfaces were simulated inside the pyroclastic
deposits and just above the ashy A soils, in agreement with the in-situ
evidence. Finally, if one or two continuous pumice soil layers are
present (schemes SL
and SE), larger portions of the hillslope are simulated as being
unstable. The role of the pumice layers can be outlined by comparing
the results obtained for the schemes QU4.5, SE4.5 and SL4.5,
respectively characterised by none, one and two pumice soil layers.
In conclusion, these analyses show that stratigraphy is an important
predisposing factor affecting the volume of the failed mass.
Finally, Fig. 13 provides the simulated failure time sequence for the
analysed schemes that significantly agrees with the in-situ evidence.
Failure conditions were first simulated for the sectors of Quindici and
then Bracigliano, Sarno-Lavorate and Sarno-Episcopio. The best fitting
of the computed depths of the slip surfaces and failure time to the in-
situ evidence was obtained by assuming c ' = 15 kPa and φ′ = 35° for
ashy soils A and c ' = 0 kPa and φ′ = 39.5° for the ashy soils B.
In addition to a satisfactory back-analysis of the May 1998 event,
the obtained results allow us to outline the role of different soil layers
on the global slope behaviour. This can be obtained by comparing
schemes with equal slope angles and depth of pyroclastic deposits but
with a different stratigraphy (SI4.5, QU4.5, SL4.5, and SE4.5). For a
homogeneous slope constituted by ashy B soils (scheme SI4.5),
rainfall and spring from bedrock cause slow reductions of FS. On the
other hand, for the scheme QU4.5 (including both A and B ashy soils),
a more complex time trend of FS is simulated. Starting from a value
similar to the previous case, FS firstly increases for 8 h and then it
rapidly decreases for the next 23 h. Different slope behaviours were
also simulated, if one or two continuous pumice soil layers are present
(SL4.5 and SE4.5). In this case, FS is almost constant for the first 30 to
39 h from 4th of May and it sharply decreases on 5th of May, from 9 to
11 p.m.
6. Conclusions
The assessment of the spatial and temporal occurrence of rainfall-
induced shallow landslides of flow type is important because of the
high risk posed by long run-out distances and high velocities of the
unstable masses. It is, however, complex when large areas are
simultaneously affected because of the heterogeneous morphological
units and stratigraphy of the landslide source areas. In order to tackle
this problem, the use of heuristic, statistical or physically based
approaches has been proposed in current literature. Heuristic or
157
statistical methods are generally suggested to evaluate the spatial
occurrence of landslides, while a joint evaluation of the spatial and
temporal occurrence can be carried out through physically based
models. However, the latter models have several limitations when
different triggering mechanisms occur in a study area. Moreover,
without heuristic analyses, the geological context of landslides is not
sufficiently analysed and misleading results may be obtained. This
paper has proposed a joint use of the different approaches and has
dealt with one of the most catastrophic landslide events in the
Campania Region (Southern Italy) which has a very high landslide
risk.
Heuristic analyses allowed us to characterise the landslide source
areas but they cannot explain the spatial occurrence of the landslides
if only the predisposing factors are considered, i.e. morphometry and
thickness/stratigraphy of slope deposits. Conversely, satisfactory
results were obtained when the triggering mechanisms were taken
into account. The triggering factors play an important role at different
scales (e.g. rainfall at massif scale and local hydraulic boundary
conditions at slope scale). They also determine the time failure
sequence. Moreover, the location and features of the source areas
significantly affect the run-out distance of the landslides. Finally,
focusing on a triggering mechanism, geomechanical analyses outline
that a low value of trasmissivity of soil deposits is an important
predisposing factor, and the stratigraphy determines the volume and
temporal occurrence of the failures.
The obtained results highlight how to assess the spatial and
temporal occurrence of shallow landslides in a large area. Heuristic
analyses can be used to identify the landslide source areas, outline the
predisposing and triggering factors of landslides as well as assess the
landslides spatial–temporal occurrence. When satisfactory results are
not obtained for the last two points, geomechanical analyses can be
carried out to correctly capture the most relevant factors. In particular,
parametric analyses can outline the role of soil properties, stratigra-
phy and hydraulic boundary conditions that affect the groundwater
regime and, consequently, the spatial and temporal occurrence of the
landslides.
References
Bilotta, E., Cascini, L., Foresta, V., Sorbino, G., 2005. Geotechnical characterization of
pyroclastic soils involved in huge flowslides. Geotechnical and Geological
Engineering 23, 365–402.
Bilotta, E., Foresta, V., Migliaro, G., 2008. The influence of suction on stiffness, shear
strength, viscosity and collapse of some volcanic ashy soils. Proc. European
Conference on Unsaturated Soils, 2–4 July, Durham UK, pp. 349–354.
Budetta, P., de Riso, R., 2004. The mobility of some debris flows in pyroclastic deposits of
the northwestern Campanian region southern Italy. Bulletin of Engineering
Geology and the Environment 63, 293–302.
158 L. Cascini et al. / Geomorphology 126 (2011) 148–158
Caine, N., 1980. The rainfall intensity-duration control of shallow landslides and debris
flows. Geografiska Annaler 62, 23–27.
Capra, L., Lugo-Hubp, J., Borselli, L., 2003. Mass movements in tropical volcanic terrains:
the case of Teziùtlan Mexico. Engineering Geology 69, 359–379.
Cascini, L., 2004. The flowslides of May 1998 in the Campania region, Italy: the scientific
emergency management. Italian Geotechnical Journal 2, 11–44.
Cascini, L., Sorbino, G., Cuomo, S., 2003. Modelling of flowslides triggering in pyroclastic
soils. Proc. Int. Conference on “Fast Slope Movements Prediction and Prevention for
Risk Mitigation”, Napoli, Patron Ed., 1, pp. 93–100.
Cascini, L., Cuomo, S., Sorbino, G., 2005. Flow-like mass movements in pyroclastic soils:
remarks on the modelling of triggering mechanisms. Italian Geotechnical Journal 4,
11–31.
Cascini, L., Guida, D., Sorbino, G., 2006. Il Presidio Territoriale: una esperienza sul
campo, G.N.D.C.I.-C.N.R Edition. 139 pp.
Cascini, L., Ferlisi, S., Vitolo, E., 2008a. Individual and societal risk owing to landslides in
the Campania region (southern Italy). Georisk: Assessment and Management of
Risk for Engineered Systems and Geohazards 2 (3), 125–140.
Cascini, L., Cuomo, S., Guida, D., 2008b. Typical source areas of May 1998 flow-like mass
movements in the Campania region, Southern Italy. Engineering Geology 96, 107–125.
Cascini, L., Cuomo, S., Pastor, M., 2008c. The role played by mountain tracks on rainfall-
induced shallow landslides: a case study. In: Sànchez-Marrè, M., Béjar, J., Comas, J.,
Rizzoli, A.E., Guariso, G. (Eds.), Proceedings of the iEMSs Fourth Biennial Meeting:
International Congress on Environmental Modelling and Software (iEMSs 2008).
7–10 July 2008, Barcelona, Spain, International Environmental Modelling and
Software Society (iEMSs), Manno, Switzerland, pp. 1484–1491.
Cascini, L., Cuomo, S., Pastor, M., Fernández-Merodo, J.A., 2008d. Geomechanical
modelling of triggering mechanisms for rainfall-induced triangular shallow
landslides of the flow-type. Proceedings of the iEMSs Fourth Biennial Meeting:
International Congress on Environmental Modelling and Software (iEMSs 2008).
7–10 July 2008, Barcelona, Spain, International Environmental Modelling and
Software Society (iEMSs), Manno, Switzerland, pp. 1516–1523.
Cascini, L., Cuomo, S., Pastor, M., Sorbino, G., 2010. Modelling of rainfall-induced shallow
landslides of the flow-type. ASCE's Journal of Geotechnical and Geoenvironmental
Engineering 1, 85–98.
Corominas, J., 1996. The angle of reach as a mobility index for small and large landslides.
Candian Geotechnical Journal 33, 260–271.
Corominas, J., Copons, R., Vilaplana, J.M., Altimir, J., Amigó, J., 2003. Integrated landslide
susceptibility and hazard assessment in the Principality of Andorra. Natural Hazards
30, 421–435.
Crosta, G.B., Dal Negro, P., 2003. Observations and modelling of soil slip-debris flow
initiation processes in pyroclastic deposits: the Sarno 1998 event. Natural Hazards
and Earth System Science 3, 53–69.
Dai, C.F., Lee, C.F., 2002. Terrain based mapping of landslide susceptibility using a geographic
information system: a case study. Canadian Geotechnical Journal 38, 911–923.
De Vita, P., 2000. Fenomeni di instabilità delle coperture piroclastiche dei Monti Lattari,
di Sarno e di Salerno Campania) ed analisi degli eventi pluviometrici determinanti.
Quaderni di Geologia Applicata 213–235.
Di Crescenzo, G., Santo, A., 2005. Debris slides-rapid earth flows in the carbonate
massifs of the Campania region Southern Italy): morphological and morphometric
data for evaluating triggering susceptibility. Geomorphology 66, 255–276.
Dietrich, W.E., Montgomery, D.R., 1998. SHALSTAB: a digital terrain model for mapping
shallow landslide potential. http://istsocrates.berkeley.edu/geomorph/ 1998.
Fredlund, D.G., Morgenstern, N.R., Widger, R.A., 1978. The shear strength of unsaturated
soils. Canadian Geotechnical Journal 15, 313–321.
Fuchu, D., Lee, C.F., Sijing, W., 1999. Analysis of rainstorm-induced slide-debris flows on
natural terrain of Lantau Island, Hong Kong. Engineering Geology 51, 279–290.
Geoslope, 2005. User's Guide. GeoStudio 2004, Version 6.13. Geo-Slope Int. Ltd., Calgary,
Canada.
Godt, J.W., Baum, R.L., Savage, W.Z., Salciarini, D., Schulz, W.H., Harp, E.L., 2008. Transient
deterministic shallow landslide modeling: requirements for susceptibility and hazard
assessments in a GIS frame work. Engineering Geology 102, 214–226.
Guadagno, F.M., Forte, R., Revellino, P., Fiorillo, F., Focareta, M., 2005. Some aspects of the
initiation of debris avalanches in the Campania Region: the role of morphological slope
discontinuities and the development of failure. Geomorphology 66, 237–254.
Hungr, O., Evans, S.G., Bovis, M.J., Hutchinson, J.N., 2001. A review of the classification of
landslides of the flow type. Environmental and Engineering Geoscience 7, 221–238.
Janbu, N., 1954. Application of composite slip surface for stability analysis. Proceedings
of the European Conference on Stability of Earth Slopes. Stockholm, Sweden, Vol. 3,
pp. 43–49.
Lacerda, W.A., 2004. The behaviour of colluvial slopes in a tropical environment. In:
Lacerda, W., Ehrlich, M., Fontoura, S.A.B., Sayão, A.S.F. (Eds.), Landslides: Evaluation
and Stabilization, Vol. 2. Balkema Publishers, London, pp. 1315–1342.
Morgenstern, N.R., Price, V.E., 1965. The analysis of the stability of general slip surfaces.
Geotechnique 15, 79–93.
Pareschi, M.T., Santacroce, R., Sulpizio, R., Zanchetta, G., 2002. Volcaniclastic debris
flows in the Clanio Valley (Campania, Italy): insights for the assessment of hazard
potential. Geomorphology 43, 219–231.
Pastor, M., Haddad, B., Sorbino, G., Cuomo, S., Drempetic, V., 2009. A depth-integrated,
coupled SPH model for flow-like landslides and related phenomena. International
Journal for Numerical and Analytical Methods in Geomechanics 33, 143–284.
Picarelli, L., Olivares, L., Andreozzi, L., Damiano, E., Lampitiello, S., 2004. A research on
rainfall-induced flowslides in unsaturated soils of pyroclastic origin. In: Lacerda,
W., Ehrlich, M., Fontoura, S.A.B., Sayão, A.S.F. (Eds.), Landslides: Evaluation and
Stabilization, Vol. 1. Balkema Publishers, London, pp. 1497–1506.
Revellino, P., Hungr, O., Guadagno, F.M., Evans, S.G., 2004. Velocity and runout
prediction of destructive debris flows and debris avalanches in pyroclastic deposits,
Campania region, Italy. Environmental Geology 45, 295–311.
Richards, L.A., 1931. Capillary conductions of liquids through porous mediums. Journal
of Applied Physics 1, 318–333.
Rossi, F., Chirico, G.B., 1998. Definizione delle soglie pluviometriche d'allarme. G.N.D.C.I.-Dept.
of Civ. Eng., University of Salerno, Italy.
Sirangelo, B., Braca, G., 2004. Identification of hazard conditions of mudflow occurrence
by hydrological model. Application of FLaIR model to Sarno warning system.
Engineering Geology 73, 267–276.
Sorbino, G., 2005. Numerical modelling of soil suction measurements in pyroclastic
soils. In: Tarantino, A., Romero, E., Cui, Y.J. (Eds.), Int. Symp. Advanced Experimental
Unsaturated Soil Mechanics. Taylor & Francis Group, London, pp. 541–547.
Sorbino, G., Foresta, V., 2002. Unsaturated hydraulic characteristics of pyroclastic soils.
Proc. 3rd International Conference on Unsaturated Soils, Recife (Brasil), 1. Balkema,
pp. 405–410.
Sorbino, G., Sica, C., Cascini, L., 2009. Susceptibility analysis of shallow landslides source
areas using physically based models. Natural Hazards 53, 313–332.
Take, W.A., Bolton, M.D., Wong, P.C.P., Yeung, F.J., 2004. Evaluation of landslide
triggering mechanisms in model fill slopes. Landslides 1, 173–184.
U.O. 2.38, 1998. Relazione sull'attività svolta dall'U.O. dell'Università di Salerno dal 21/
05/1998 al 5/07/1998. G.N.D.C.I.-Dept. of Civil Eng., University of Salerno.
Van Westen, C.J., 2004. Geo-Information tools for landslide risk assessment: an overview of
recent developments. In: Lacerda, W., Ehrlich, M., Fontoura, S.A.B., Sayão, A.S.F. (Eds.),
Landslides: Evaluation and Stabilization, Vol. 2. Balkema Publishers, London, pp. 39–56.
Wang, F.W., Sassa, K., Wang, G., 2002. Mechanism of a long-runout landslide triggered
by the August 1998 heavy rainfall in Fukushima Prefecture, Japan. Engineering
Geology 63, 169–185.