Rainfall Thresholds for Possible Occurrence

of Shallow Landslides and Debris Flows in Italy

Maria Teresa Brunetti, Fabio Luino, Carmela Vennari, Silvia Peruccacci,

Marcella Biddoccu, Daniela Valigi, Silvia Luciani, Chiara Giorgia Cirio,
Mauro Rossi, Guido Nigrelli, Francesca Ardizzone, Mara Di Palma,
and Fausto Guzzetti

1 Introduction

In mountain regions worldwide, rainfall-induced landslides and associated debris

flows erode slopes, scour channels, and contribute to the formation of alluvial fans
that may harm humans and destroy buildings. Rainfall-induced slope failures are
frequent and widespread in Italy, where individual rainfall events can result in
single or multiple slope failures in small areas or in very large regions. Most of
the harmful failures were rainfall-induced, and several were shallow slides or debris
flows. In the 60-year period 1950–2009, casualties due to landslides were at least
6,349, an average of 16 harmful events per annum. The large number of harmful
events indicates the considerable risk posed by rainfall-induced shallow landslides
and debris flows to the population of Italy (Guzzetti et al. 2005a; Salvati et al. 2010).
In this work, we exploit information on rainfall events that have resulted in
shallow failures and debris flows to define regional thresholds for the possible
occurrence of rainfall-induced shallow landslides and debris flows in three regions in
northern, central, and southern Italy. Following a review of methods for establishing
the dependence of landslide occurrence on rainfall measurements, we illustrate three
catalogues of rainfall events that have resulted in slope failures in the Abruzzo,
Calabria and Lombardy regions. Next, we present the method used for the definition

M.T. Brunetti () • S. Peruccacci • S. Luciani • M. Rossi • F. Ardizzone • F. Guzzetti

CNR IRPI, via Madonna Alta 126, 06128 Perugia, Italy
e-mail: [email protected]
F. Luino • M. Biddoccu • C.G. Cirio • G. Nigrelli • M.D. Palma
CNR IRPI, Strada delle Cacce 73, 10135 Torino, Italy
e-mail: [email protected]; [email protected]
C. Vennari • D. Valigi
Dipartimento di Scienze della Terra, Università degli Studi di Perugia,
piazza dell’Università 1, 06123 Perugia, Italy

M. Schneuwly-Bollschweiler et al. (eds.), Dating Torrential Processes on Fans and 327

Cones, Advances in Global Change Research 47, DOI 10.1007/978-94-007-4336-6 22,
© Springer ScienceCBusiness Media Dordrecht 2013
328 M.T. Brunetti et al.

of objective rainfall thresholds, and we apply the method to the available catalogues
to determine new intensity-duration (ID) thresholds for possible occurrence of
shallow landslides and debris flows in the three regions. We conclude by comparing
the new thresholds to similar empirical thresholds proposed for possible landslide
initiation in Italy.

2 Modelling Approaches

Two approaches have been proposed to evaluate the dependence of landslide

occurrence (or lack of occurrence) on rainfall measurements (Reichenbach et al.
1998; Guzzetti et al. 2007). The first approach relies on process-based models
(Montgomery and Dietrich 1994; Wilson and Wieczorek 1995; Wu and Sidle
1995; Iverson 2000; Crosta and Frattini 2003), and the second approach is based
on empirical thresholds (Caine 1980; Reichenbach et al. 1998; Corominas 2000;
Aleotti 2004; Wieczorek and Glade 2005; Guzzetti et al. 2007).
Process-based models rely upon the understanding – and the schematization – of
the physical laws controlling slope instability, and attempt to extend spatially the
simplified stability models widely adopted in geotechnical engineering. Stability
conditions are evaluated chiefly by means of a static stability model where the local
equilibrium of forces along a pre-defined slip surface is considered. Most com-
monly, the slip surface is assumed planar, at a fixed depth, and parallel to the topo-
graphic surface. Values for the pore fluid pressure are assumed, or obtained adopting
rainfall infiltration models, coupled to surface and sub-surface runoff models.
The main problem of process-based models consists in the difficulty of obtaining
reliable and accurate information necessary to construct and calibrate the distributed
models. The problem is particularly severe where the models are applied to large
areas. The simplest model requires eight parameters (i.e., terrain gradient, soil
density, friction angle, cohesion, hydraulic conductivity, failure depth, water table
depth, and rainfall history), of which only terrain gradient is known reasonably
accurately over large areas through the use of digital terrain models. Although a
stochastic approach may partly solve the problem of insufficient geotechnical and
hydrological data, lack of distributed information constrains the applicability of
process-based models to areas of limited extent, and mostly to the back analysis
of individual rainfall events. This reduces the use of process-based models in
operational landslide forecasting systems (Aleotti 2004). A second limitation of
process-based models is that, for a given rainfall event, the models predict as
unstable an area significantly larger than the area actually affected by the event
landslides. The overestimation of the extent of the unstable areas has multiple
reasons that must be considered when the model outputs are used in landslide
forecasting systems. A different method consists in using process-based models to
determine the amount of rainfall (i.e., the critical rainfall) that, when and where it
infiltrates in the ground, can result in slope instability, and use this information to
forecast possible landslide occurrence during a rainfall event, or using quantitative
rainfall forecasts or estimates (Chang et al. 2007; Chiang and Chang 2008).
Rainfall Thresholds for Possible Occurrence of Shallow Landslides. . . 329

The second approach to relate landslide occurrence on rainfall measurements

consists in the use of empirical thresholds. An empirical threshold defines the
rainfall, soil moisture, or hydrological conditions that, when reached or exceeded,
are likely to result in landslides (Crozier 1996; Reichenbach et al. 1998). Rainfall
thresholds are the most common type of empirical thresholds for possible landslide
occurrence, and are defined through the analysis of past rainfall events that have
(or have not) resulted in slope failures. Rainfall thresholds are commonly classified
on the geographical extent for which they are determined, and the type of rainfall
information used to establish the threshold (Guzzetti et al. 2007).
Based on the geographical extent, rainfall thresholds can be loosely subdivided
as global, national, regional, or local thresholds. A global threshold attempts to
establish a general (“worldwide”) minimum level below which landslides do not
occur, independently of local morphological, lithological and land-use conditions
and of local or regional rainfall pattern and history. Global thresholds were proposed
by Caine (1980), Innes (1983), Clarizia et al. (1998), Crosta and Frattini (2001),
Cannon and Gartner (2005), Hong et al. (2006) and Guzzetti et al. (2008). National
thresholds are similar to global thresholds, and are established using information
for a single nation (e.g., Brunetti et al. 2010). Regional thresholds are defined
for areas extending from a few to several thousand square kilometres of similar
meteorological, climatic, physiographic, and soil characteristics. Local thresholds
consider the local climatic regime and geomorphic setting, and are applicable to
single landslides or to groups of landslides in areas extending from a few to some
hundreds of square kilometres.
Based on the rainfall measurements, empirical rainfall thresholds can be grouped
in three broad categories: (i) thresholds that combine precipitation measurements
obtained for specific rainfall events, (ii) thresholds that include the antecedent
rainfall and soil moisture conditions (e.g., Terlien 1998; Crozier 1999; Glade 2000;
Chleborad 2003; Aleotti 2004), and (iii) other thresholds (e.g., Jakob and Weatherly
2003). Thresholds established using precipitation measurements obtained from indi-
vidual or multiple rainfall events can be further subdivided into: intensity–duration
thresholds, thresholds based on the total event rainfall, rainfall event–duration
thresholds, and rainfall event–intensity thresholds (for a review see Guzzetti et al.
2007, 2008). To account for meteorological and climate influence on slope stability,
rainfall measurements (chiefly rainfall intensity) can be normalized to measures of
climate, including the mean annual precipitation, the average number of rainy-days,
and the rainy-days normal (Wilson and Jayko 1997).

3 Landslide and Rainfall Information

Information on rainfall events that have resulted in landslides exists in Italy (see
e.g. Crosta 1998; Crosta and Frattini 2001; Aleotti 2004; Guzzetti et al. 2007,
2008; Brunetti et al. 2010, and references therein). For this work, we considered the
existing information and added new information to obtain three regional catalogues
330 M.T. Brunetti et al.

Table 1 Characteristics of the regional catalogues used to establish empirical rainfall thresholds
for possible occurrence of shallow landslides and debris flows
Lombardy Abruzzo Calabria
Number of rainfall 173 85 113
events
Period (year) 9/1929–7/2008 8/2002–10/2009 1/1996–2/2010
Rainfall duration (h) 1 < D < 91 1 < D < 600 2 < D < 1080
Rainfall mean intensity 0.6 < I < 32 0.3 < I < 50 0.3 < I < 29
(mm h1 )
Number of rain gauges 56 25 65
Area extent (km2 ) 8,000 11,000 15,000
Mean area per rain 140 430 230
gauge (km2 )
Elevation range (m) 150–3,769 0–2,914 0–2,267
Geographical location Northern Italy Central Italy Southern Italy
Topography Mountains Hills, mountains Hills, mountains
Climate Mountain, cold Mediterranean, mild Mediterranean, warm

of rainfall events that have resulted in shallow landslides and debris flows, for the
Lombardy region, northern Italy, for the Abruzzo region, central Italy, and for the
Calabria region, southern Italy (Table 1).
Luino et al. (2008) searched historical archives, old chronicles, journals, and
regional and local newspapers to obtain a comprehensive catalogue of rainfall
induced soil slips and debris flows for the Sondrio and Brescia Provinces, Lombardy
region. Systematic analysis of the historical information resulted in a catalogue of
661 individual landslides, in the 80-year period 1929–2008. The location and time
of occurrence of the failures were obtained from the historical sources and the press.
Newspapers were most useful to decide the timing of the failures, whereas the exact
or approximate geographical coordinates of the failures were obtained in a GIS
using large-scale digital base maps. Once the failures were located geographically,
rainfall information for the individual landslides was obtained from a database
of rainfall measurements made available by the Lombardy region. The database
contained rainfall data for 56 rain gauges in the Brescia and Sondrio provinces,
corresponding to an average density of one rain gauge every 140 km2 . For each
single landslide, selection of the most appropriate rain gauge was based on: (i) the
location of the rain gauge, preferably placed in the same catchment as the landslide,
(ii) the distance of the rain gauge to the landslide, not to exceed 5 km, and (iii) the
availability of sufficient rainfall data for the considered landslide event.
The catalogue obtained by Luino et al. (2008) listed individual soil slips and
debris flows caused by rainfall. For consistency with the information available
for the Abruzzo (Brunetti et al. 2010) and the Calabria regions (see below), we
transformed the list of 611 rainfall-induced landslides in a new catalogue of 160
rainfall events that have resulted in individual or multiple soil slips and debris
flows in the Brescia and Sondrio provinces. During the process, landslides listed
in separate records in the original catalogue that were triggered by an individual
Rainfall Thresholds for Possible Occurrence of Shallow Landslides. . . 331

rainfall event, and for which the rainfall information was obtained from the same
rain gauge, were merged in a single entry in the new catalogue of rainfall events
with landslides. We completed the catalogue for the Lombardy region by adding
13 events, including 4 events in the Bergamo province and 9 events in the Sondrio
province, obtained from a global catalogue of rainfall events with shallow landslides
and debris flows prepared by Guzzetti et al. (2008). Thus, the final catalogue for the
Lombardy region listed 173 rainfall events with shallow slope instabilities in an
Alpine area of about 8,000 km2 .
For the Abruzzo region, central Italy, Brunetti et al. (2010) obtained a catalogue
of 77 rainfall events with landslides in the 8-year period 2002–2009. The landslide
information was obtained through the systematic analysis of regional and local
newspapers. The exact or approximate location of the single landslides (or of
groups of landslides triggered by the same rainfall event) was decided based on
the historical information. The corresponding geographical coordinates for the
landslides were obtained using Google Earth. Analysis of the historical sources
revealed that most of the landslides in the catalogue were shallow, and small in
size. For this work, we updated this regional catalogue searching new landslide
information, and excluding landslides that could not be classified reasonably as
shallow failures or debris flows. As a result, we obtained a new catalogue listing
85 rainfall events that have resulted in shallow landslides and debris flows in the
Abruzzo region, in the period from August 2002 to October 2009.
Adopting a similar procedure, we searched more than 25 national, regional and
local newspapers and internet resources for the Calabria region, southern Italy,
and we obtained information on 191 individual rainfall events that have caused
landslides, including 113 events that have resulted in shallow failures and debris
flows, in the period from January 1996 to February 2010. Again, the timing and the
location of the individual failures (or of groups of failures triggered by the same
rainfall event) were decided through the analysis of the historical information, and
the geographical coordinates for the landslides were obtained using Google Earth.
For the Abruzzo and the Calabria regions, rainfall information was obtained from
a national database of rainfall measurements managed by the national Department
for Civil Protection. The database contains rainfall measurements for 1950 rain
gauges in Italy, including 25 rain gauges in the Abruzzo region and 65 rain gauges in
the Calabria region. This corresponds to an average density of one rain gauge every
430 km2 for Abruzzo region, and one rain gauge every 230 km2 for Calabria. In
the database, rainfall measurements are cumulated every 10, 15, 30 or 60 min, allow-
ing for a detailed reconstruction of the individual rainfall events that have resulted in
slope instabilities, in terms of rainfall duration D and rainfall mean intensity I. For
Calabria, additional rainfall information was obtained from an on-line catalogue
provided by the Regional Environmental Agency (http://www.arpacalabria.it/). For
each individual landslide, up to four rain gauges were considered, and the most
representative station was selected. Representativeness of a rain gauge was decided
based on the geographical distance to the failure, the elevation of the rain gauge
compared to the elevation of the landslide, and the location of a rain gauge with
respect to the local topographical and morphological setting. In general, the selected
332 M.T. Brunetti et al.

rain gauge was the closest to the landslide, in the set of rain gauges located in the
same valley. In other cases, for the selection of the representative rain gauge, eleva-
tion was considered more important than geographical distance to the landslide.
Table 1 summarizes the main characteristics of the three regional catalogues of
rainfall events with shallow landslides and debris flows.

4 Definition of Rainfall Thresholds

Guzzetti et al. (2007) identified the need for objective methods to define repro-
ducible thresholds of possible landslide occurrence from empirical rainfall data, and
proposed a method based on Bayesian inference. Brunetti et al. (2010) adopted a
Frequentist statistical approach, and proposed a different method for the definition
of objective rainfall thresholds. In this work, we adopt the Frequentist approach to
determine new regional rainfall thresholds for possible landslide occurrence in the
Lombardy, Abruzzo, and Calabria regions.
To define a threshold, we start by plotting the empirical rainfall duration D and
mean rainfall intensity I data in a single graph, and we find the best-fit line to
the cloud of empirical data points (D, I). As an example, Fig. 1a shows the 173
(D, I) data points for the Lombardy region. To avoid problems associated with
the fitting of data spanning multiple orders of magnitude (e.g., the least square
minimization criteria may not work), we log-transform the empirical data, and we
fit the distribution of the rainfall conditions log(I) vs. log (D) that have resulted in
slope failures with a linear equation of the type log.I / D log./  ˇ  log.D/, which
is entirely equivalent to the power law I D   D ˇ in linear coordinates, where D is
the duration of the rainfall event (in h), I is the rainfall mean intensity (in mmh1),
 is the intercept, and ˇ is the gradient of the power law curve. In Fig. 1a, the solid
line is the best-fit line TL,50 : I D 10:93  D 0:42 .
Next, for each rainfall event (D, I), we calculate the difference ı(D) between the
logarithm of the event intensity log[I(D)] and the corresponding intensity value of
the fit log[If (D)], ı(D) D log[I(D)]  log[If (D)]. Then, we estimate the probability
density of the distribution of ı, p(ı), using a kernel density estimation (KDE)
technique (Venables and Ripley 2002), and we model the distribution through least
square fitting using the Gaussian function
" #
1 .ı  /2
G.ı/ D p  exp 
2 2 2 2

where  is the mean and  is the standard deviation. In Fig. 1d, the dashed line
shows the empirical distribution (KDE) of the ı values for the Lombardy rainfall
data, and the continuous line is the equivalent Gaussian model.
Based on the modelled distribution of ı, multiple thresholds can be defined, cor-
responding to different probabilities of exceeding thresholds. As an example, Fig. 1g
portrays the 5% rainfall thresholds TL,5 for the Lombardy region. The distance
Rainfall Thresholds for Possible Occurrence of Shallow Landslides. . . 333

Fig. 1 Empirical rainfall (D, I) data, analysis, and new ID rainfall thresholds for the Lombardy
(left column), Abruzzo (central column), and Calabria (right column) regions, Italy. Plots in the
upper row show rainfall conditions that have resulted in shallow landslide and debris flows (dots),
and the associated power law best-fit lines, TL,50 for Lombardy, TA,50 for Abruzzo, and TC,50 for
Calabria. Inset maps show geographical location of each region. Graphs in the central row show the
kernel density estimation of the differences, ı, for 173 rainfall events in Lombardy, for 85 events
in Abruzzo, and for 113 events in Calabria. Solid lines are the corresponding Gaussian model fits.
Vertical lines for TL,5 , TA,5 , and TC,5 are the thresholds that correspond to 5% probability of having
landslides, for Lombardy, Abruzzo, and Calabria, respectively. Plots in lower row portray rainfall
(D, I) conditions that have resulted in shallow landslide and debris flows (dots) in the three regions.
Solid lines are the 1% (TL,1 , TA,1 , TC,1 ) and the 5% (TL,5 , TA,5 , TC,5 ) power law thresholds
334 M.T. Brunetti et al.

ı L,5 between TL,5 and TL,50 is used to calculate the intercept of the 5% threshold
curve. The 5% rainfall threshold TL,5 is the curve parallel to the best-fit line TL,50
(slope D ˇ), with intercept  L,5 D  L,50 ı L,5 . For the TL,5 threshold, assuming the
catalogue of rainfall events is sufficiently complete and representative for the Alpine
portion of the Lombardy region, we can state that the probability of experiencing
landslides triggered by rainfall below this threshold is less than 5%. The method
can be used to determine thresholds for any probability of exceeding a threshold
level (Brunetti et al. 2010).

5 New Regional Thresholds for Shallow Landslide

Occurrence

We have exploited the catalogues of rainfall events that have resulted in shallow
landslides and debris flows in the Lombardy, Abruzzo, and Calabria regions
presented in Sect. 3, and the statistical method discussed in Sect. 4, to determine new
rainfall thresholds for possible occurrence of shallow failures in the three regions.
Results are summarized in Fig. 1.
In this figure, the three upper plots show the distributions of the rainfall
conditions (D, I) that have resulted in shallow landslides and debris flows in the
Lombardy region (Fig. 1a), in the Abruzzo region (Fig. 1b), and in the Calabria
region (Fig. 1c). In the plots, the black lines are the best-fit lines to the empirical
rainfall data points, obtained through linear fitting (least square method) of the log-
transformed rainfall data [log(D), log(I)].
The three central plots in Fig. 1 show, for each region, the empirical distribution
of ı (dashed lines) and the corresponding modelled Gaussian fit (solid lines). The
spread of the modelled Gaussian distribution around the central value,  D 0, is
measured by the standard deviation , which ranges between 0.23 (for Lombardy
and Abruzzo) and 0.25 (for Calabria). The empirical distributions of the differences
ı for the three regional data sets and the corresponding Gaussian models exhibit
interesting similarities. First, the peaks of the empirical distributions are shifted
to the left of the corresponding modelled distributions. We attribute the shift,
which is largest for Calabria and reduced for Lombardy, to heterogeneities in the
data (e.g., different lithological, morphological, and climatic conditions in each
region). Second, the modelled distributions overestimate slightly the distribution
of the empirical data, for ı  0. We attribute this chiefly to the reduced number of
data points in the individual catalogues. Third, the empirical distributions exhibit a
visible deviation from the Gaussian models for ı  0.5. Inspection of the catalogues
revealed that this corresponds primarily to short duration – high intensity rainfall
conditions, which are somewhat different from the majority of the other (D, I)
conditions that have resulted in slope failures. Inspection of the catalogue revealed
that the short duration – high intensity rainfall data pertain to a single regional
meteorological event that has resulted in landslides in multiple sites in a large region,
for which several rain gauges were available.
Rainfall Thresholds for Possible Occurrence of Shallow Landslides. . . 335

In Fig. 1, the three lower plots portray the empirical rainfall thresholds, for the
three regions. The slopes of the power law curves representing the thresholds are
similar, but not identical. The slope of the threshold defined for the Lombardy
region (Fig. 1g), an area characterized by a mountain (cold) climate, is significantly
lower (ˇ D 0.42) than the thresholds defined for the Abruzzo (ˇ D 0.58, Fig. 1h)
and the Calabria (ˇ D 0.57, Fig. 1i) regions, characterized by a Mediterranean
climate. This is in agreement with previous findings by Guzzetti et al. (2007),
and it indicates that in mountain areas mean rainfall intensity is more important
than rainfall duration to decide when landslides occur, an important information to
construct an operational landslide forecasting system (Brunetti et al. 2009). Further
inspection of the threshold curves for the three regions reveals that, for the same
short rainfall duration (D < 7 h), the mean rainfall intensity I required to initiate
shallow landslides and debris flows, corresponding to 1% probability of having
landslides, is lowest in the Lombardy region (Fig. 1g, D D 2 h, I D 2.39 mmh1),
which is characterized by a mountain climate, intermediate in the Abruzzo region
(Fig. 1h, D D 2 h, I D 2.87 mm h1 ) that exhibits a mild Mediterranean climate, and
largest in Calabria (Fig. 1i, D D 2 h, I D 3.56 mmh1), where the climate is warm
Mediterranean. Conversely, for long rainfall duration (D > 30 h), the mean rainfall
intensity required to initiate shallow slope failures (at the 1% probability level) is
lowest in Abruzzo, intermediate in Calabria, and largest in Lombardy. We conclude
that in Italy, the rainfall conditions (D, I) that can result in shallow landslides and
debris flows depend on the prevailing meteorological conditions and climate. The
two are related, and conditioned on latitude and the physiographical setting.

6 Discussion and Conclusions

The new regional rainfall ID thresholds for possible occurrence of shallow land-
slides and debris flows established for the Lombardy, Abruzzo, and Calabria regions
can be compared to similar empirical thresholds proposed in Italy for individual
regions, single catchments, or local areas. Figure 2 portrays six regional (thick lines)
and 11 local (thin lines) ID rainfall thresholds proposed in Italy, including our three
new thresholds TL,1 (for Lombardy), TA,1 (for Abruzzo), and TC,1 (for Calabria).
Inspection of Fig. 2 indicates that the new regional thresholds are significantly lower
than the regional thresholds previously proposed for Piedmont (Aleotti 2004), for
Lombardy (Ceriani et al. 1994), and for the Campania region (Calcaterra et al.
2000). Our new Lombardy threshold, defined using rainfall data primarily for
the Brescia and Sondrio provinces, is also significantly lower than the threshold
proposed by Cancelli and Nova (1985) for Valtellina, a portion of the Sondrio
province. Further, our new regional thresholds are significantly lower than the
thresholds proposed for individual catchments in the Susa Valley, Piedmont (Bolley
and Oliaro 1999), for the Moscardo torrent, Friuli (Marchi et al. 2002), for the
Apuane Alps, Tuscany (Giannecchini 2005), and for the Valzangona area, Marche
(Floris et al. 2004).
336 M.T. Brunetti et al.

Fig. 2 Comparison of regional and local rainfall thresholds for possible landslide occurrence in
Italy. Regional (thick line) and local (thin line) thresholds, and new 1% probability of exceeding
the thresholds for the Abruzzo (TA,1 ), Calabria (TC,1 ), and Lombardy (TL,1 ) regions (Source: 1,
Ceriani et al. 1994; 2, Calcaterra et al. 2000; 3, Aleotti 2004; 4, Cancelli and Nova 1985; 5–10,
Bolley and Olliaro 1999; 11, Marchi et al. 2002; 12, Floris et al. 2004; 13–14, Giannecchini 2005)

The fact that the new thresholds are systematically lower than previously
established thresholds, even for the same general area (e.g., for Lombardy (Ceriani
et al. 1994), or Valtellina (Cancelli and Nova 1985)) deserves an explanation. Even if
we cannot exclude that the result is – at least partially – attributable to local climatic,
meteorological, morphological, lithological, and land cover settings, we maintain
that it depends primarily on the type and quality of the information used to define the
thresholds. Most often, thresholds published in the literature are defined following
major (i.e., extreme, catastrophic, rare) rainfall events that have resulted in abundant
and widespread landsliding. With this respect, many of the published thresholds
are – or are close to – “maximum” thresholds, i.e. thresholds that represent rainfall
conditions above which there is a 100% chance of landslide occurrence (Crozier
1996; Reichenbach et al. 1998). Conversely, our thresholds are defined from
catalogues that include low intensity (i.e., non catastrophic, more frequent) rainfall
events that have resulted in single, or sparse landslides. Further, our catalogues
are the result of a systematic and thorough analysis of the landslide and rainfall
information available for the study areas. This has resulted in more complete lists
of rainfall events with associated slope failures, which have provided better data
sets for the definition of empirical thresholds based on statistical methods. We take
Rainfall Thresholds for Possible Occurrence of Shallow Landslides. . . 337

this as an indication of the need for the systematic collection of reliable, accurate,
and as complete as reasonably possible information on rainfall events that have
resulted in individual or multiple slope failures, to be able to determine dependable
rainfall thresholds for possible landslide initiation to be used in landslide forecasting
systems (Brunetti et al. 2009).
The finding that in the areas investigated in this study, and possibly in other
areas in Italy, shallow slope failures and associated debris flows can be triggered by
less severe rainfall conditions than previously recognized has multiple implications.
First, it is important for the design and the implementation of landslide forecasting
systems (Brunetti et al. 2009, 2010). Second, it indicates that the temporal frequency
of shallow failures is higher than previously considered, with consequences for haz-
ard assessment (Guzzetti et al. 2005b, 2006) and risk evaluation. Third, it suggests
that ongoing and foreseen climate and environmental changes, including human
induced changes, can have a higher impact on landslide activity than expected.

Acknowledgements Part of the paper is reprinted from the work of Brunetti et al. (2010), and it
is used with permission. Work supported by the Italian National Department for Civil Protection
(DPC), which provided the national rainfall database for the period 2002–2009. MTB, SP and MR
were supported by DPC grants. We thank G. Tonelli for the work on the rainfall database, and E.
E. Brabb for reviewing the paper.

References

Aleotti P (2004) A warning system for rainfall-induced shallow failures. Eng Geol 73:247–265
Bolley S, Oliaro P (1999) Analisi dei debris flows in alcuni bacini campione dell’Alta Val Susa.
Geoingegneria Ambientale e Mineraria, Marzo, 69–74
Brunetti MT, Peruccacci S, Rossi M, Guzzetti F, Reichenbach P, Ardizzone F, Cardinali M,
Mondini A, Salvati P, Tonelli G, Valigi D, Luciani S (2009) A prototype system to forecast
rainfall induced landslides in Italy. In: Picarelli L, Tommasi P, Urciuoli G, Versace P (eds)
Rainfall-induced landslides: mechanisms, monitoring techniques and nowcasting models for
early warning systems, vol 1. Naples, pp 157–161
Brunetti MT, Peruccacci S, Rossi M, Luciani S, Valigi D, Guzzetti F (2010) Rainfall thresholds for
the possible occurrence of landslides in Italy. Nat Hazard Earth Syst Sci 10:447–458
Caine N (1980) The rainfall intensity-duration control of shallow landslides and debris flow. Geogr
Ann Ser A 62:23–27
Calcaterra D, Parise M, Palma B, Pelella L (2000) The influence of meteoric events in triggering
shallow landslides in pyroclastic deposits of campania, Italy. In: Bromhead E, Dixon N, Ibsen
ML (eds) Proceedings of the 8th international symposium on landslides, vol 1. A.A. Balkema
Publishers, Cardiff, pp 209–214
Cancelli A, Nova R (1985) Landslides in soil debris cover triggered by rainstorms in Valtellina
(central Alps – Italy). In: Proceedings of the 4th international conference and field workshop
on landslides. The Japan Geological Society, Tokyo, pp 267–272
Cannon SH, Gartner JE (2005) Wildfire-related debris flow from a hazards perspective. In: Jakob
M, Hungr O (eds) Debris flow hazards and related phenomena. Springer, Berlin, pp 363–385
Ceriani M, Lauzi S, Padovan N (1994) Rainfall thresholds triggering debris-flows in the alpine
area of Lombardia Region, central Alps – Italy. Man and Mountain ’94 Proceedings, Ponte di
Legno, pp 123–139
338 M.T. Brunetti et al.

Chang K, Chiang SH, Lei F (2007) Analysing the relationship between typhoon-triggered
landslides and critical rainfall conditions. Earth Surf Process Landforms 33(8):1261–1271
Chiang SH, Chang K (2008) Application of radar data to modelling rainfall-induced landslides.
Geomorphology 103(3):299–309
Chleborad AF (2003) Preliminary evaluation of a precipitation threshold for anticipating the
occurrence of landslides in the Seattle, Washington, DC. US geological survey open-file report:
03–463 F
Clarizia M, Gullà G, Sorbino G (1998) Sui meccanismi di innesco dei soil slip. In: Proceedings
of International conference on prevention of hydro-geological hazards: The role of scientific
research. Alba, Italy, 5–7 Nov 1996, vol 1, pp 585–597
Corominas J (2000) Landslides and climate. Keynote lecture. In: Bromhead E, Dixon N, Ibsen
ML (eds) Proceedings of the 8th international symposium on landslides, vol 4. A.A. Balkema
Publishers, Cardiff, pp 1–33
Crosta GB (1998) Regionalization of rainfall thresholds: an aid to landslide hazard evaluation.
Environ Geol 35(2–3):131–145
Crosta GB, Frattini P (2001) Rainfall thresholds for triggering soil slips and debris flow. In: Mugnai
A, Guzzetti F, Roth G (eds) Proceedings of the 2nd EGS Plinius conference on Mediterranean
storms. Siena, Italy, 16–18 Oct 2001, pp 463–487
Crosta GB, Frattini P (2003) Distributed modelling of shallow landslides triggered by intense
rainfall. Nat Hazard Earth Syst Sci 3(1–2):81–93
Crozier MJ (1996) The climate-landslide couple: a southern hemisphere perspective. Paleoclim
Res 19 ESF(Special Issue 12):329–350
Crozier MJ (1999) Prediction of rainfall-triggered landslides: a test of the antecedent water status
model. Earth Surf Process Landforms 24:825–833
Floris M, Mari M, Romeo RW, Gori U (2004) Modelling of landslide-triggering factors – a case
study in the Northern Apennines, Italy. Lect Note Earth Sci 104:745–753
Giannecchini R (2005) Rainfall triggering soil slips in the southern Apuan Alps (Tuscany, Italy).
Adv Geosci 2:21–24
Glade T (2000) Modelling landslide triggering rainfall thresholds at a range of complexities. In:
Bromhead E, Dixon N, Ibsen ML (eds) Landslides in research, theory and practice. Thomas
Telford, Cardiff, pp 633–640
Guzzetti F, Stark CP, Salvati P (2005a) Evaluation of flood and landslide risk to the population of
Italy. Environ Manage 36(1):15–36
Guzzetti F, Reichenbach P, Cardinali M, Galli M, Ardizzone F (2005b) Probabilistic landslide
hazard assessment at the basin scale. Geomorphology 72:272–299
Guzzetti F, Galli M, Reichenbach P, Ardizzone F, Cardinali M (2006) Landslide hazard assessment
in the Collazzone area, Umbria, central Italy. Nat Hazard Earth Syst Sci 6:115–131
Guzzetti F, Peruccacci S, Rossi M, Stark CP (2007) Rainfall thresholds for the initiation of
landslides in central and southern Europe. Meteorol Atmos Phys 98:239–267
Guzzetti F, Peruccacci S, Rossi M, Stark CP (2008) The rainfall intensity-duration control of
shallow landslides and debris flows: an update. Landslides 5(1):3–17
Hong Y, Adler R, Huffman G (2006) Evaluation of the potential of NASA multi-satellite
precipitation analysis in global landslide hazard assessment. Geophys Res Lett 33:L22402
Innes JL (1983) Debris flows. Prog Phys Geogr 7:469–501
Iverson RM (2000) Landslide triggering by rain infiltration. Water Resour Res 36(7):1897–1910
Jakob M, Weatherly H (2003) A hydroclimatic threshold for landslide initiation on the North Shore
Mountains of Vancouver, British Columbia. Geomorphology 54:137–156
Luino F, Biddoccu M, Cirio CG, Nigrelli G, Di Palma M (2008) Definizione delle soglie
pluviometriche d’innesco di frane superficiali e colate torrentizie: accorpamento per aree
omogenee. Quaderni IReR, monografia, rapporto finale della convenzione fra CNR IRPI e
Protezione Civile Regione Lombardia, 125 pp
Marchi L, Arattano M, Deganutti AM (2002) Ten years of debris-flow monitoring in the Moscardo
Torrent (Italian Alps). Geomorphology 46:1–17
Rainfall Thresholds for Possible Occurrence of Shallow Landslides. . . 339

Montgomery DR, Dietrich WE (1994) A physically based model for the topographic control of
shallow landsliding. Water Resour Res 30(4):1153–1171
Reichenbach P, Cardinali M, De Vita P, Guzzetti F (1998) Regional hydrological thresholds for
landslides and floods in the Tiber River Basin (Central Italy). Environ Geol 35(2–3):146–159
Salvati P, Bianchi C, Rossi M, Guzzetti F (2010) Societal landslide and flood risk in Italy. Nat
Hazard Earth Syst Sci 10:465–483
Terlien MTJ (1998) The determination of statistical and deterministic hydrological landslide-
triggering thresholds. Environ Geol 35(2–3):124–130
Venables WN, Ripley BD (2002) Modern applied statistics with S. Springer, New York
Wieczorek GF, Glade T (2005) Climatic factors influencing occurrence of debris flows. In: Jakob
M, Hungr O (eds) Debris flow hazards and related phenomena. Springer, Berlin/Heidelberg,
pp 325–362
Wilson RC, Jayko AS (1997) Preliminary maps showing rainfall thresholds for debris-flow activity,
San Francisco Bay Region, California. US Geological survey open-file report 97–745 F
Wilson RC, Wieczorek GF (1995) Rainfall thresholds for the initiation of debris flow at La Honda,
California. Environ Eng Geosci 1(1):11–27
Wu W, Sidle RC (1995) A distributed slope stability model for steep forested basins. Water Resour
Res 31:2097–2110