Remote sensing as tool for development of landslide databases: The case
of the Messina Province (Italy) geodatabase
Andrea Ciampalini a,
⁎, Federico Raspini a
, Silvia Bianchini a
, William Frodella a
, Federica Bardi a
,
Daniela Lagomarsino a
, Federico Di Traglia a
, Sandro Moretti a
, Chiara Proietti b
, Paola Pagliara b
, Roberta Onori b
,
Angelo Corazza b
, Andrea Duro b
, Giuseppe Basile c
, Nicola Casagli a
a
Department of Earth Science, University of Firenze, via La Pira 4, Firenze 50121, Italy
b
Italian Civil Protection Department (DPC), via Vitorchiano 2, Roma 00189, Italy
c
Regional Department of Civil Protection (DRPC), Hydrogeological and Environmental Risks Service and Service for Messina Province, Italy
a b s t r a c ta r t i c l e i n f o
Article history:
Received 27 August 2014
Received in revised form 21 January 2015
Accepted 23 January 2015
Available online 3 February 2015
Keywords:
Geodatabase
Landslide
Remote sensing
Risk
Civil Protection
Hazard
Landslide geodatabases, including inventories and thematic data, today are fundamental tools for national and/or
local authorities in susceptibility, hazard and risk management. A well organized landslide geo-database contains
different kinds of data such as past information (landslide inventory maps), ancillary data and updated remote
sensing (space-borne and ground based) data, which can be integrated in order to produce landslide susceptibility
maps, updated landslide inventory maps and hazard and risk assessment maps. Italy is strongly affected by
landslide phenomena which cause victims and significant economic damage to buildings and infrastructure,
loss of productive soils and pasture lands. In particular, the Messina Province (southern Italy) represents an
area where landslides are recurrent and characterized by high magnitude, due to several predisposing factors
(e.g. morphology, land use, lithologies) and different triggering mechanisms (meteorological conditions,
seismicity, active tectonics and volcanic activity). For this area, a geodatabase was created by using different
monitoring techniques, including remote sensing (e.g. SAR satellite ERS1/2, ENVISAT, RADARSAT-1, TerraSARX,
COSMO-SkyMed) data, and in situ measurements (e.g. GBInSAR, damage assessment). In this paper a complete
landslide geodatabase of the Messina Province, designed following the requirements of the local and national
Civil Protection authorities, is presented. This geo-database was used to produce maps (e.g. susceptibility,
ground deformation velocities, damage assessment, risk zonation) which today are constantly used by the
Civil Protection authorities to manage the landslide hazard of the Messina Province.
© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
Geodatabases dedicated to landslides, including inventories and
thematic data, represent a powerful tool for users operating at different
administrative and organizational levels, from local to continental
(Guzzetti et al., 1994; Dellow et al., 2003; Gaspar et al., 2004; Colombo
et al., 2005; Devoli et al., 2007; Gustavvson et al., 2008; Suprechi et al.,
2010; Van den Eeckhaut and Hervás, 2012; Battistini et al., 2013;
Klose et al., 2014; Pereira et al., 2014; Zêzere et al., 2014). The potential
of a geodatabase is given by past information (landslide inventory maps,
LIM) combined with geoenvironmental and triggering factors, in order
to produce useful maps (e.g. landslide susceptibility map, LSM, map of
the element at risk) and to manage the different emergency cycle
phases (van Westen et al., 2005). Furthermore all of these data and
products, organized into a GIS database, can be easily updated when
new acquired data are included. GIS technologies facilitate the management
of different typology of analogical data, maintaining a proper cartographic
representation (Guzzetti et al., 2012) and allowing the
combination of analogue and vector data in order to create useful hazard
management maps. These different information, included in the
GIS geodatabase maintain their geometric consistency, allowing rapid
query and calculation using the stored data (e.g. landslide area,
typology, hazard degree).
In Italy, landslides are recurrent phenomena causing victims and
significant economic damage to buildings and infrastructures, loss of
productive soils and pasture lands. Annually, during the rainy seasons
(from September to November and from March to June) the Italian
regions are affected by natural hazards, among which landslides represent
the primary cause of death (Guzzetti, 2000).
The Sicily Region (southern Italy) and in particular the Messina
Province is an area prone to landslide hazard due to several predisposing
factors (morphology, land use, lithologies) and of different triggering
mechanisms as meteorological conditions (Del Ventisette et al., 2012),
Geomorphology 249 (2015) 103–118
⁎ Corresponding author. Tel.: +39 0552757548; fax: +39 0552756323.
E-mail address: andrea.ciampalini@unifi.it (A. Ciampalini).
http://dx.doi.org/10.1016/j.geomorph.2015.01.029
0169-555X/© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Contents lists available at ScienceDirect
Geomorphology
journal homepage: www.elsevier.com/locate/geomorph
seismicity, active tectonics (Billi et al., 2008) and volcanic activity
(Intrieri et al., 2013; Nolesini et al., 2013). For instance, between 2009
and 2010 in both the Messina Province mountain ranges (Nebrodi
and Peloritani), several municipalities were affected by more than a
thousand different types of mass movements (i.e. debris flows, complex,
rotational and translational landslides, rock falls). These landslides
caused more than 37 victims and severe damage to urbanized areas,
infrastructures (national highways, rural roads and railroads), cultivated
and pasture lands. The blocking of rural roads left several towns isolated
causing serious difficulties in the rescue activities. Some of these events
occurred in areas characterized by low to null landslide hazard, affecting
urbanized areas considered as safe, suggesting that the previously available
landslide information and mapping underestimated the potential
hazard of this area. From the beginning of this emergency phase the
Civil Protection System was promptly activated, in order to manage
the landslide risk, involving the Civil Protection Departments at both
National (DPCN) and Regional (DPRC) levels, the Environmental
Department and other local authorities and research centers (Pagliara
et al., 2014).
In the framework of the post-disaster phase led by the DPCN, several
activities were performed aimed at supporting local authorities in the
hydrogeological risk assessment and management. Following the
above mentioned events the Messina Province drew the attention of
many geoscientists, becoming test site of the FP7 European projects
DORIS (www.doris-project.eu) and LAMPRE (www.lampre-project.
eu). Within these projects all the previously available data (thematic
maps, Digital Elevation Model, VHR optical photos) and newly acquired
data (optical and Synthetic Aperture Radar, SAR), were integrated into
an homogenized GIS geodatabase. All these data were used in order to
improve the products (e.g. landslide inventory maps, ground deformation
velocity maps, residual risk zonation), and their value added in
landslide mapping, monitoring and forecasting activities.
In the Messina Province, a geodatabase was used to include different
products created by means of advanced monitoring techniques,
comprising remote sensing (optical and SAR) and in situ (Ground BasedInterferometric
SAR, GBInSAR) data. In particular, remotely sensed data acquired
through different SAR satellites (ERS1/2, ENVISAT, RADARSAT-1,
TerraSAR-X, COSMO-SkyMed) were used in order to monitor the long
term slope deformation evolution. The long time displacement series obtained
by combining ERS1/2 and ENVISAT data (1992–2010) and different
sensors (C- and X- band) allowed to update the landslide state of activity
and to refine the extension of single landslides by comparing the ground
deformation velocities measured in different periods.
In this paper the structures and contents of the Messina Province
landslide geodatabase are described. Furthermore, some very useful
products at different scales generated by the combination of LIMs,
thematic data and remote sensing methodologies are presented,
highlighting their use by the Civil Protection authorities.
2. Study area
The study area (Messina Province) is located in southern Italy, in
Sicily's North-Eastern sector (Fig. 1). From a geologic point of view the
Messina Province is part of the Sicilian fold-and-thrust orogenic belt
formed during the Tertiary due to the convergence between the
European and African plates (Dewey et al., 1989; De Guidi et al.,
2013). This belt is made of three different tectono–stratigraphic zones
overthrust on a buried structural domain (Di Paolo et al., 2014). The
internal orogenic sector corresponds to the Peloritani Mountains
which are made of crystalline rocks and a Meso–Cenozoic sedimentary
cover (Messina et al., 2004). The Central Zone (Sicilidi Units) represents
the accretionary wedge (upper Cretaceous–lower Miocene pelagic
succession) formed by the remnants of the Neotethyan wedge
(Corrado et al., 2009). The External Zone (upper Oligocene–middle
Miocene clastic sediments) is represented by the Numidian Flysch
Units (Carbone et al., 1990; Thomas et al., 2010). Rocks belonging to
the Central and External Zones crop out within the Nebrodi Mountains.
From North-East to South-West several wedge-top deposits (Oligocene–
Pliocene) can be recognized (Di Paolo et al., 2014). Along the Tyrrhenian
and Ionian coasts several normal faults formed small extensional basins
during the Pleistocene, filled by terrigenous deposits (Di Stefano et al.,
2012).
The geostructural conditions, the competence of the outcropping
rocks and the tectonic activity strongly influence the morphology
of the area. The coastal landscape is typical for recently uplifted areas:
steep slopes, narrow valleys and high relief energy are the main
geomorphologic features. The morphometric characteristics of the
river basins are represented by straight and parallel hydrographic
networks influenced by the short distance separating the watershed
from the coast. River catchments have a narrow with significant transport
of solid materials; incisions are short and deeply entrenched into
V-shaped valleys, especially in the mountainous sectors. Several small
alluvial plains, formed where riverbeds become over-flooded, characterize
the coastal area.
Forming a hydrographic point of view, the presence of straight, steep
course, gravel-bed river draining mountain areas is the most evident
feature. Their presence is typical for the Mediterranean climate region
and their flow varies seasonally and their regime is torrential with
catastrophic transport of solid materials following heavy rainfall, causing
severe damage if flooding occurs close to densely populated centers.
Intense and exceptional rainfall events were the main triggering
factors which, combined with predisposing factors such as weak
lithologies and the steep slopes, caused several mass movements in
the Messina Province, such as: debris flow, rotational and translational
slides, rock falls and shallow and deep-seated landslides.
2.1. The event of autumn 2009
On October 1st 2009, a high intensity storm hit the Ionian side of
Sicily, particularly affecting a restricted area in the south-western part
of Messina Province, between the Peloritani Mountains ridge and the
coastline (Fig. 1) (Ardizzone et al., 2012; Del Ventisette et al., 2012). In
this zone, the main built-up areas are concentrated along the coastline.
The hilly area is characterized by the presence of smaller settlements
(e.g. Briga, Giampilieri, Altolia) which are connected with the urbanized
coastal area by a single provincial road climbing upward along very
steep slopes.
The St. Stefano di Briga rain gauge (Fig. 1) recorded a rainfall of
225 mm in 8 h, with a peak of 115 mm in 3 h. The heavy precipitation
of October 1st further reduced the slope stability conditions, already
undermined by the significant accumulation of rainfall that had
occurred during the month of September, which led to complete soil
saturation. During the same day, the persistent rainfall triggered more
than 600 slope failures, mainly shallow soil slides, debris avalanches,
and debris and mud flows over an area of about 50 km2
. The more
severe damage was reported in the villages of Giampilieri, Scaletta
Zanclea, Guidomandri, Pèzzolo, Altolìa, and Itàla. Landslides damaged
many roads as well as the railway system and main highways. The
assessed number of fatalities caused by landslides and inundation
amounted to 37 (including 31 deaths and 6 missing persons), with
122 injured people and 2019 evacuated people.
2.2. The event of winter 2010
Between 2009 and 2010 several municipalities (San Fratello,
Caronia, Naso, Ucria, and many others) located in the Nebrodi
Mountains, a 70 km long ENE-WSW trending mountain chain, spanning
in altitude from the sea level to 1847 m (Fig. 1), were intensely affected
by several slope instability phenomena that caused intense damages to
buildings and infrastructures (Del Ventisette et al., 2013; Bardi et al.,
2014; Bianchini et al., 2014a,b; Ciampalini et al., 2014). This area located
in the western part of the Messina Province was affected mainly by
104 A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
complex, rotational and deep-seated phenomena. The deformation velocity,
slower with respect to those of the debris flows that occurred
in the Peloritani Mountains autumn 2009 event, prevented the presence
of victims but the damage to buildings and infrastructures was severe.
Following this event, in the framework of the post-disaster
initiatives lead by the National Department of Civil Protection, several
activities were performed aimed at supporting local authorities in the
hydrogeological risk assessment. In particular, a detailed geomorphologic
analysis, exploiting the contribution of PSInSAR™
technique and
photo interpretation, was performed by the Earth Science Department
of the University of Florence (DST-UNIFI, Centre of Competence of the
Italian Civil Protection for geo-hazards) in order to identify slow moving
pre-existing landslides affecting the whole territory hit by a heavy
rainfall period.
3. Methods
The database of the Messina Province was developed in three different
phases: (i) data collection; (ii) database design; and (iii) product
generation. The collected data can be divided into two categories:
(i) ancillary data and (ii) space-borne SAR data (Fig. 2).
3.1. Data collection
3.1.1. Ancillary data
Ancillary data included in the database are represented by different
thematic maps (geological, topographical, building and infrastructure
maps, land cover, and pre-existing landslide inventory maps, Digital
Elevation Model, VHR optical photos). The geology of the study area
was retrieved using the geological map of the Messina Province (scale
1:50,000) in digital format. The land cover was deduced by the updated
(2000) Corine Land Cover map (scale 1:100,000). Each polygon was
classified following the three levels of the Corine Land Cover classification
(http://www.eea.europa.eu/publications/COR0-landcover). The
C.T.R. (Carta Tecnica Regionale, scale 1:10,000) in vector format was
used for the topography and for the localization of buildings and
infrastructures. Slope and aspect maps of the Messina Province were
produced using a DEM with a 20 × 20 m resolution generated by the
Fig. 1. (a) Location of the study area. Dots represent the municipalities affected by landslide phenomena between 2009 and 2010. (b) Geolithological map of the Messina Province.
(See online version for color images).
105A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
Istituto Geografico Militare (IGM). An orthophoto (Volo Italia 2000, 1 m
of pixel resolution) of the Sicily Region was used as reference map.
Quickbird VHR-optical images were also used for specific areas. Usually,
these images were acquired before and after major investigated events
such as those occurred in Giamplieri and in San Fratello between October
2009 and February 2010.
The landslide inventory map of the Sicily Region (PAI — Piano
di Assetto Idrogeologico) was produced between 2005 and 2007
and contains information about 32,669 slope instability phenomena
distributed within the Sicily Region. Among the mapped landslides,
9358 phenomena (28.65%) are located within the Messina Province. A
smaller part of the inventory (about the 5%) is updated up to 2012.
Each landslide is represented by a polygon which is linked to an
attribute table where the related available data (ID, hydrographic
basin, province, municipality, typology, state of activity, hazard degree,
perimeter, area) are reported.
The database includes very detailed information regarding specific
municipalities damaged by severe events during the last five years. For
example internal reports of the Regional Civil Protection Department
are available for several municipalities (e.g. Caronia, San Fratello, Naso,
Castell'Umberto). These reports contain detailed information related to
single landslides (inclinometers, geophysical investigations, boreholes,
fracture patterns, building and infrastructure damage assessment
maps).
In particular for the San Fratello Municipality ground displacement
maps are also available. They were obtained using a ground based
radar which started to acquire data since March 10th 2010.
3.1.2. Space-borne SAR data
Ground displacement measurements were obtained for the Nebrodi
and Peloritani mountain ranges using the SqueeSAR™
technique
(Ferretti et al., 2011). Developed by T.R.E. (Tele-Rilevamento Europa, a
spin-off company of the Polytechnic University of Milano), the
SqueeSAR approach is a second generation PSInSAR algorithm (Ferretti
et al., 2000, 2001), one of many different approaches (Raucoules et al.,
2009) based on the processing of several (at least 15), co-registered,
multi-temporal satellite SAR images of the same target area. The main
idea behind the PSInSAR method is to analyze the backscattered signal
from the observed scene to identify ground elements (temporally stable
from an electromagnetic point of view) characterized by high reflectivity
(Persistent Scatterers, PSs) (Ferretti et al., 2000, 2001; Werner et al.,
2003). A single PS exhibits reduced temporal decorrelation due to its
high electromagnetic reflectivity and its coherent, point-like, stable
scattering behavior. Having a stable radar signature, PSs are only slightly
affected by decorrelation phenomena and the level of backscattered
signal is much higher than the inherent noise of the sensor (i.e., the
signal-to-noise ratio is extremely high).
Fig. 2. Flow chart showing the design of the Messina Province geodatabase and how the single layers were used and combined in product generation. Data were divided between available
ancillary data and new acquired space-borne SAR data. (See online version for color image).
106 A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
The development of the SqueeSAR method (Ferretti et al., 2011)
allows a step ahead in the conventional processing techniques and has
proven to successfully overcome the main limitations of the pixelbased
analysis. In fact, in natural, less urbanized and agricultural environments,
PSInSAR technique partially fails due to complete
decorrelation of a great number of scatterers. SqueeSAR technique provides
a significantly increased coverage of ground deformation measurement
points, especially over non-urban areas (Meisina et al.,
2013; Bellotti et al., 2014), detecting not only point-wise Permanent
Scatterers (as the PSInSAR technique) but also spatially Distributed
Scatterers (DSs). This challenging new approach is capable of obtaining
reliable ground displacement measurement, retrieving electromagnetic
information not only from point-wise, corner reflector-like objects
(i.e., PS), but also from DSs, identified in low-reflectivity homogeneous
areas, such as scattered outcropping rocks, debris flows, noncultivated
lands and arid areas.
A number of satellite SAR datasets were acquired and processed to
produce ground deformation maps for the investigated areas. SAR
images acquired by C-band (ERS 1/2, ENVISAT and RADARSAT-1) and
by X-band (COSMO-SkyMed, CSK and TerraSAR-X) were collected and
processed (Table 1). Processed SAR data include both pre-event images
(for the mapping of pre-existing extremely slow to very slow moving
landslides) and post-event information (exploited to assess residual
movements and deformation).
An attribute table is associated to each interferometric dataset,
including: (i) PS identifier (ID); (ii) geographic coordinates (Latitude,
Longitude); (iii) height; (iv) average velocity and the related standard
deviation; (v) coherence degree and (vi) the cumulative displacement
for each acquisition date, expressed in millimeters. The latter fields are
used to retrieve time displacement series.
3.2. Database design
The database was created between 2010 and 2013, within the
European Union's Seventh Framework Programme for Research Project
“Advanced downstream service for the detection, mapping, monitoring
and forecasting of ground deformations” (FP7 DORIS Project). The aim
was to provide geo-information products designed for the detection,
mapping, monitoring and forecasting of ground deformations, with emphasis
on products useful to the Regional Civil Protection Department
(Del Ventisette et al., 2013). All geographic information was projected
using the WGS 84 projection datum and the Universal Transversal
Mercator as coordinate system.
3.3. Product generation
The products delivered using the geodatabase were divided into two
different categories: (i) basic products and (ii) advanced products. The
former group includes products created using a single category of data
(ancillary data or space-borne SAR data), whereas advanced products
were generated combining the two categories. The quality of the
space-borne data was assessed before the product generation step in
order to evaluate their usefulness and limitations.
3.3.1. Quality assessment of PS data
The PS detection suitability of the used PS datasets was evaluated
considering the satellite parameters and the morphology of the study
area, following the approach developed by Notti et al. (2010, 2014),
which is a useful method to analyze the geometrical visibility of the
area of interest. The suitability was evaluated by using two different
indexes: the R-index (RI) and the land use (LU) index. The former
represents the ratio between the slant range and the ground range
(Notti et al., 2014) and provides the suitability of a given area to be
investigated by a certain satellite system, taking into account the
satellite parameters (incident angle) as well as the morphological
characteristics of the slope (angle and orientation) derived from the
available DEM. The LU index depends both on the land use and on the
satellite sensor and PSI processing strategies (Notti et al., 2014). The RI
ranges from the maximum value +1 to negative value. It assumes the
maximum value when the suitability of the area to be investigated
through the selected PS dataset is the highest. This means that the
slope direction of the area is approximately parallel to the satellite LOS
direction. On the contrary, the R-index assumes negative values, or
around 0, when geometrical distortion effects prevent an acceptable
visibility of the area. Considering the LU index and simplifying the
land coverage of the test area, five land cover classes were defined:
(i) urban built-up zones, (ii) rocky and debris areas, (iii) crops,
(iv) forests and (v) water reservoirs.
3.3.2. Basic product generation
Part of the collected ancillary data such as the geological maps, land
cover map, the pre-existing landslide inventory and the DEM derived
products were used to produce the landslide susceptibility map (LSM)
of the Messina Province. A simple implementation of Random Forest, a
machine learning algorithm developed by Breiman (2001) that
performs a multivariate classification, was developed in a Matlab
environment in order to produce the LSM. The complete description of
the methodology applied to LSM generation is reported in Catani et al.
(2013). Numerous parameters were considered, in order to avoid the
subjectivity in the choice of explanatory variables: elevation, curvature
(general, planar, profile), slope, aspect, flow accumulation, Topographic
Wetness Index (TWI), land use and lithology (Fig. 2). The source data
are represented by a cell of 20 × 20 m: the average value inside a
100 × 100 m cell was calculated. For each parameter the standard
deviation for numeric variables, and the variety for the categorical
ones, were also calculated; always inside a 100 × 100 m window
considering the variability of 25 cells (Lagomarsino et al., 2014).
Ground deformation velocity maps were obtained by using each
interferometric dataset. In particular these maps consider the velocity
measured along the LOS and, usually, they require an independent
ground truth data validation (in situ measurements or observations).
Following the methodology proposed in Notti et al. (2014) a stability
Table 1
Characteristics of the PSI datasets collected in the geo-database of the Messina Province.
Satellite Area Geometry Temporal range N° of images N° of PS PS/km2
ERS 1/2 Nebrodi Ascending 23/08/1992–13/12/2000 138 43,405 23.22
ERS 1/2 Nebrodi Descending 01/05/1992–08/01/2001 140 104,754 24.35
ENVISAT Nebrodi Ascending 22/01/2003–11/10/2010 107 137,770 38.07
ENVISAT Nebrodi Descending 07/07/2003–13/09/2010 98 99,175 32.46
RADARSAT-1 Nebrodi Ascending 30/12/2005–26/01/2010 46 225,465 112.73
RADARSAT-1 Nebrodi Descending 31/01/2006–03/02/2010 47 107,326 85.86
CSK Nebrodi Ascending 01/05/2011–03/05/2012 51 453,331 288.31
CSK Nebrodi Descending 16/05/2011–02/05/2012 64 793,725 729.94
TSX Nebrodi Descending 28/10/2011–22/09/2012 90 551,091 379.67
ERS 1/2 Peloritani Ascending 08/09/1992–21/11/2000 48 6607 96.2
ENVISAT Peloritani Ascending 22/01/2003–20/05/2009 55 30,411 414.9
107A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
threshold (±2 mm/yr) was used for the available C-band data. A more
precautionary threshold was chosen for the X-band data (±5 mm/yr)
because of their higher standard deviation due to the shorter acquisition
period (Table 1). These maps can be used at a regional scale.
Combining and integrating ground deformation velocity maps
extracted by means of SqueeSAR technique with optical data
(orthophoto, VHR optical satellite images and multi-temporal aerial
photo), it was possible to highlight the presence of areas affected by
high hydrogeological risk (hotspot mapping) at the local scale
(Raspini et al., 2013).
3.3.3. Advanced product generation
Displacements and ground deformation velocities are measured by
the satellite in the direction of the LOS whereas the real displacement
actually occurs in three dimensions (Cascini et al., 2009). For this reason,
a method to obtain a more reliable measured velocity is a key issue in
landslide monitoring. The measured PS LOS velocities are projected
along the steepest slope direction. The displacement is referred to the
direction of the local maximum slope gradient, under the assumption
that it is the most probable direction of real movement for a potential
landslide phenomenon (Cascini et al., 2010; Notti et al., 2010; Plank
et al., 2012; Bianchini et al., 2013; Herrera et al., 2013). In order to
calculate the velocity projected along the slope (VSlope mm/yr), the
formula proposed by Bianchini et al. (2013) and Notti et al. (2014)
was adopted. This formula considers: (i) the slope and aspect derived
from the DEM; (ii) and the satellite acquisition parameters (i.e. the
azimuth, the incidence angle, the directional cosines: nlos, hlos, and
elos of the LOS). Thus, all the displacement velocities measured along
the LOS (VLos) of all the available PS datasets were projected along the
steepest slope, assumed (VSlope) as the most probable direction of
movement. The VSlope velocity range is usually larger than VLos, since
the minimum detected rate and the standard deviation are higher
because of the major dispersion of the PS population. This method is
useful when information acquired by different satellite sensors (with
different LOS) and different acquisition orbits (i.e. ascending and
descending) are available, in order to easily interpret and compare
displacements projected along the same direction (steepest slope
direction).
PS dataset can be profitably used to update an existing landslide
inventory map. In particular, due to the acquisition parameters of the
satellite systems (i.e. signal wavelength, revisiting time), this technique
can only be applied to slow-moving landslide phenomena (Ciampalini
et al., 2012) which according to Cruden and Varnes (1996) are
classified as extremely slow (velocity b 16 mm/yr) and very slow
(16 mm/yr b velocity b 1.6 m/yr). Therefore those phenomena classified
as “rapid erosion” and “rapid flow”, due to their high velocity, were
excluded from the pre-existing database statistics analysis since they
cannot be detected by satellite systems. Moreover, very small landslides
(i.e., extension b 1 ha) were also removed from the database before
computing statistics. It is in fact almost impossible to have a minimum
required number of 4 PS (Righini et al., 2012) located within a single
landslide having such limited extension. The updating procedure was
performed using a landslide activity matrix proposed by Righini et al.
(2012). The state of activity assessment is based on the comparison
between the mean velocity detected by the older available dataset and
the one obtained using the more recent dataset.
PS datasets were combined with building maps at a municipality in
order to produce building deformation velocity maps following the
procedure proposed in Ciampalini et al. (2014). The deformation velocity
of each building was calculated considering the average velocity of
all the PS included in the building polygon. All the available C- and
X-band datasets were used to monitor the evolution of the building
deformation between the 1992 and 2012. A stability threshold of the
building deformation between ±1.5 mm/yr was chosen for the C-band
data and ± 2 mm/yr for the X-band data.
The integration between space-borne and ground based SAR
data was applied at a local (slope) scale in the Nebrodi area, for highly
accurate monitoring and mapping activities on the landslide affecting
the town of San Fratello, one of the most severely damaged urban centers
after the winter 2010 event (Bardi et al., 2014). The GBInSAR
employed system was designed and implemented by the Joint Research
Centre (JRC) of the European Commission and its spin-off company
Ellegi-LiSALab (Tarchi et al., 2003a,b; Casagli et al., 2009, 2010; Gigli
et al., 2011; Di Traglia et al., 2014a,b). The radar system was installed
at an average distance of 2100 m from the landslide. The area covered
by a GBInSAR system depends on the distance between the sensor and
the target, which is usually limited to a few hundreds of meters up to
a few kilometers. In the specific case of San Fratello, the covered area
is about 1 km2
, approximately corresponding to the landslide extension
(Bardi et al., 2014). Cumulative displacement maps were obtained by
means of a GBInSAR system dedicated to the monitoring of the February
14th 2010 landslide. The monitoring campaign was carried out from
March 10th 2010 to March 14th 2013. Ground displacements were
measured along the GBInSAR line of sight (LOS). The precision in the
displacement measurement is 0.1 mm for punctual (few pixels) automatic
extraction and displacement time series are obtained through
the time differentiation of displacements obtained from two consecutive
images (Di Traglia et al., 2014b). The integration of PS and GBInSAR
approaches allows to overcome the physical limitations of each single
technique, providing the displacement analysis of the same area from
different points of view and in different temporal intervals.
Space-borne interferometric data combined with geological, geomorphological
and hydrological data were used to refine the residual
risk zonation mapping in areas affected by the 2009 event debris
flows. Risk zonation was performed for built up areas using four
different risk levels (Table 2) considering the proximity of past debris
flow run out areas and of the riverbeds, building damages and presence
of active slow moving landslides detected using PS data.
Table 2
Different levels of risk adopted for the risk zonation.
Risk level Description
Level 0 – Green Negligible geomorphological and hydraulic risk. Areas not affected by landslides.
Population allowed to come back to their houses after the verification of the existence of access road and the functionality of services and basic
infrastructure (water, electricity, sewage, etc.).
Level 1 – Yellow Low geomorphological and hydraulic risk. Areas affected by landslides or areas nearby zones affected by landslides.
Buildings can be used with some caution. Limited geotechnical and geomorphological stabilization works are required. Limited hydraulic mitigation
works have to be performed. Civil Protection plans are required.
Level 2 – Pink Medium to high geomorphological and hydraulic risk. Areas affected by landslides or areas nearby zones affected by landslides.
Buildings can be used only after extensive stabilization works.
Presence of buildings and infrastructures that may increase the geomorphological and hydraulic hazard.
Building demolition and relocation cannot be excluded.
Level 3 – Red Very high geomorphological and hydraulic risk
Areas close or within riverbed. Areas within landslides. In these areas buildings must be demolished to allow the restoration of pre-existing stream.
In these areas, the structures (e.g., buildings) and/or infrastructure (e.g., roads) will be relocated to safer areas.
108 A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
4. Results
4.1. Quality assessment of PS data
The R-Index was calculated for all the available radar datasets
and classified into 5 classes. A good agreement between R-Index and
PS density could be noticed: the better the R-Index is, the greater
amount of PS and the greater PS density were detected. A general
improvement of the RI from C-band sensors to X-band one was
observed, but among the C-band sensors RADARSAT-1 shows higher
value with respect to ERS1/2 and ENVISAT (Fig. 3). The inversion
between the highest and lowest RI values related to the orientation of
the slopes was observed in all the datasets acquired both in ascending
and descending geometry (ERS 1/2, ENVISAT, RADARSAT-1 and CSK).
Thus, when comparing ascending and descending passes of the same
satellite, it can be observed that the highest and the lowest R-Index
classes are basically inverted on slopes facing E and W (Fig. 3). It is
worth highlight that the Messina Province area is particularly suitable
Fig. 3. RI distribution for the Messina Province considering the acquisition parameters (incident angle) of each satellite and the morphology of the study area (slope, aspect). The RI ranges
from the +1 to negative value assuming the maximum value when the morphology of the area is optimal to be investigated through the selected PS dataset. (See online version for color
images).
109A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
for this relief visibility analysis because it is characterized by the presence
of N–S oriented valleys with E or W facing slopes. The enhancement
of the RI can be related to the differences in the incident angle,
since the greater is the incidence angle, the greater is the R-Index. The
average PS density is higher for the X-band sensors because of their
higher resolution, which allows the detection of a greater amount of
radar benchmarks. ERS 1/2 and ENVISAT show the higher number of
PS in the 0.25–0.50 class, whereas RADARSAT-1 and X-band sensors
(CSK and TSX) reveal a greater PS amount in classes characterized by
higher value of the R-Index.
The evaluation of L-index deals with the PS number on land use,
which depends on the types of land coverage of the area and on their
characteristics of maintaining the scattering properties in time, as well
as on the diverse microwave band data acquired by different satellite
sensors. Areas with buildings and structures are characterized by
the highest PS densities. In these areas the presence of human made
structures, corresponding to the most stable reflectors, increases the
PS density for both the acquisition bands (C- and X-bands). Large rock
and talus outcrops can be considered as good stable reflectors too,
showing good PS densities, especially in X-band. Water reservoir areas
show high PS density because they correspond to seasonal streams
which are dry for most part of the year. During the dry season, the
river beds are formed by blocks corresponding to stable PS.
4.2. Basic products
4.2.1. Landslide susceptibility map
The landslide susceptibility map of the Messina Province (Fig. 4)
shows that most of the analyzed area is affected by moderate to very
high landslide susceptibility. Most of the areas scarcely affected by
ground deformation phenomena are located along the coast, in correspondence
with the narrow coastal alluvial plains. The map also shows
that the Nebrodi Mountains are more susceptible than the Peloritani
Mountains. The performance of the model used was evaluated by
building a ROC (Receiver Operating Characteristic) curve, shown in
Fig. 4. The AUC (Area Under Curve) value of 0.73 highlights a good
result.
4.2.2. Ground deformation velocity maps
Ground deformation velocity maps (Fig. 5) can be used at regional
scale in order to detect the most unstable areas. These maps can be
elaborated clustering unstable and stable PS in order to determine
active and stable areas. This kind of map is very useful in regional
studies because most unstable areas can be easily and immediately
detected allowing the attention to be focussed on specific targets. The
comparison among PS dataset acquired in different time intervals can
be used to evaluate the evolution in time of the areas affected by slow
moving ground deformation phenomena.
4.2.3. Giampilieri area hot spot mapping
The abovementioned approach can be performed also at a local
scale. Following the autumn 2009 event previously described, a multitemporal
analysis of satellite radar images, processed through the
SqueeSAR algorithm, was exploited for detecting and mapping slope
instability at basin scale in the south-western part of the Messina
Province. Available PS data provided estimates of yearly deformation
velocity, referred to both historical (1992–2001; ERS 1/2 data) and
recent (2003–2009; ENVISAT data) scenarios. The investigated area,
with an extent of about 75 km2
, is located in a narrow coastal area
south of the suburbs of the city of Messina, delimited by the Ionian sea
on the East side and by the Peloritani ridge on the West side. The area
includes the following five municipal towns, located in the Messina
district: Alì, Alì Terme, Itala, Scaletta Zanclea and Messina. The hotspot
mapping procedure identified 26 sites (Fig. 6) characterized by high
hydrogeological risk (Raspini et al., 2013). On the basis of available
multi–interferometric data, these sites were assessed as the most
critical in terms of hydrogeological hazard and risk, for the type and
extent of the detected landslide phenomena, for the measured deformation
velocities and for the presence of elements at risk (Fig. 6). For each
site, a PS landslide mapping was performed and validated through field
surveys, and the updating of the pre-existing landslide inventory. A
validated hotspot analysis (Altolia village, site number 14 in Fig. 6),
representative of the type phenomena identified in the whole study
area, is presented. Fig. 7 shows the results from photo and radar interpretation
for Altolia, a village in the Messina municipality with 460
inhabitants. The settlement has developed along the right bank of
Fig. 4. Landslide susceptibility map derived from the available ancillary data (elevation, curvature, (general, planar, profile), slope, aspect, flow accumulation, TWI, land use and lithology)
included in the Messina Province geo-database and the related ROC curve. The susceptibility map was produced using a simple implementation of Random Forest. (See online version for
color image).
110 A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
Giampilieri river. An overview of the SqueeSAR results, expressed in
mm/yr, both for ERS 1/2 and ENVISAT datasets is shown in Fig. 7a and
b. The PS results highlighted ground motion displacements in both
historical (1992–2001; ERS 1/2) and recent (2003–2009; ENVISAT)
scenarios. In particular, available datasets allowed the identification
of a large sector in the southeastern part of the village characterized
by displacements with an average velocity ranging from 2 mm/yr to
10.7 mm/yr in the ERS 1/2 dataset, and from 2.2 mm/yr to 8.9 mm/yr
in the ENVISAT dataset. Photo interpretation of stereoscopic color
images (1:3500 scale) and field surveys allowed the detection and
mapping of a large active continuous slide affecting the south-eastern
part of Altolia village (Fig. 7c). During the field campaign, carried out
in the framework of the post-disaster activities and aimed at validating
SqueeSAR results, several effects on infrastructure (building cracks and
damage) were detected. A detailed study of landform features (scarps,
counterscarps, tension cracks) led to a complete characterization
of the Altolia landslide, which could be divided in different sectors
characterized by homogenous morphologic behavior (crown and
accumulation zone).
4.3. Advanced products
4.3.1. VSlope ground deformation velocity maps
A more accurate evaluation of the deformation velocities at the local
scale was performed converting LOS velocities into VSlope velocities
combining the ground deformation velocity maps and the Messina
Province 20 × 20 m DEM. The VLos projection along the local slope
leads to a significant reduction of the PS population (Notti et al.,
2014): PS amount with VSlope velocities was reduced for ERS 1/2 dataset
to the 38% of the VLos PS amount. For ENVISAT dataset, PS population
was reduced to the 42%. VSlope RADARSAT-1 is the 43% of the PS VLos.
Within CSK dataset, VSlope dataset is the 22% of VLos population. TSX PS
VSlope were the 84% of the VLos population. This loss of information in
terms of PS amount was compensated by the more reliable evaluation
of the ground deformation using the VSlope instead of the VLos. According
to the combination of the local topography (including aspect and
gradient of the slopes) with the acquisition geometry of the satellite,
some movements within landslide boundaries seemed to be uplift
motions since they revealed a movement toward the satellite sensor.
This misunderstanding interpretation can be partially solved by conversion
to VSlope values, that allows observing the movement in agreement
with the real expected displacement down the slope, and thus permits a
more reliable and robust analysis of the slope dynamics.
4.3.2. Updated landslide inventory map
The updated landslide inventory map (LIM), obtained by combining
the pre-existing landslide inventory (PAI, 2012) and PS datasets, was
performed in a test area (25 km2
) located in the western part of the
Messina Province. The PAI is dated up to 2012 and includes landslide
phenomena classified with respect to the typology and the state of
activity (active, dormant, inactive – including relict and abandoned
phenomena – and stabilized), according to a simplified version of
Cruden and Varnes (1996) classification. The updated LIM was realized
using the following satellite SAR datasets, RADARSAT-1 (2005–2010)
and CSK (2011–2012), both acquired in ascending and descending
geometries. The procedure applied allowed the identification of new
landslides, modifying their boundaries and assessing the state of activity
following the Cruden and Varnes (1996) classification. The assessment
of the state of activity was based on the comparison between the
mean velocity detected by the older available dataset (RADARSAT-1 in
this case) and that obtained using the more recent one (CSK). The
results obtained showed an increase in the accuracy of the updated
Fig. 5. Example of ground deformation velocity maps obtained through (A) ascending ERS1/2 (1992–2001), (B) ascending ENVISAT (2003–2010). The area characterized by a cluster of
blue PS in (A) corresponds to the northern sector of the Etna volcano. (See online version for color images).
111A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
LIM with 25 new detected phenomena (13.16%) and the enlargement of
other 35 landslides (18.42%).
4.3.3. Building deformation velocity map (San Fratello)
Buildings are sensitive to movements caused by landslides. The
assessment of the building deformation in landslide area, using the PS
datasets, represents a useful tool in order to understand the landslide
evolution, magnitude and stress distribution. This method can
be applied both in a municipality affected by past landslides, monitoring
residual movements, and in other areas detected as unstable, by means
of the hotspot mapping, but not yet involved in slope instability
phenomena. Furthermore, the building deformation velocity maps
obtained, which are valuable products to monitor the safety of
urban areas affected by ground deformation, can be easily validated
through dedicated field surveys devoted to the detection and evaluation
of building damage. The designed geodatabase was used to
monitor the evolution of the building deformations that occurred
between the 1992 (ERS1/2) and the 2012 (CSK) in the San Fratello
Municipality, generating different building deformation velocity maps.
The maps produced (Fig. 8) clearly show an increase in the number of
buildings affected by deformation phenomena measured after the
1992 but before the event of the large landslide that occurred on
February 14th 2010. The pre-event data highlight deformation also
along the western slope of the town, affected by a historical landslide
that occurred in 1922, detecting slow but continuous deformation. The
most interesting results were obtained using the CSK dataset acquired
during the post event phase. The related building deformation velocity
map highlights the presence of important residual deformations in the
areas most intensely damaged during the 2010 landslide (Fig. 8b).
4.3.4. Local deformation analysis (San Fratello)
In the town of San Fratello a detailed local deformation analysis
was performed combining space-borne and GBInSAR data (Fig. 9).
Cumulative displacement maps were used to identify areas characterized
by different ground motion. The peak deformations were located
in correspondence with the western landslide boundary (Fig. 10).
Ground deformation in these areas was evaluated also using the space
borne data acquired along different LOS with respect to the GBInSAR
system.
The C-band data acquired between 1992 and 2010 were used to
evaluate the potential risk of the area. X-band and GBInSAR data were
acquired to monitor ground deformation during the post event phase.
The combined space borne (both C- and X-band) and GBInSAR analysis
was performed in order to refine the landslide boundary. With respect
to the landslide boundary, assessed by means of a geomorphological
field survey carried out by the Sicily Region Civil Protection Department
Fig. 6. Location of the 26 sites characterized by high hydro-geological hazard (hotspot mapping). 1) M. Santo; 2) P.le Puzzu; 3) Alì Terme; 4) P.le Seddiri; 5) P.le Cimmario; 6) P.le Pitarra;
7) Culma Caravagi; 8) Itàla; 9) Croce; 10) Itàla Marina; 11) M. Pietralunga; 12) P.le Lanzaro; 13) P.zzo Scapola; 14) Altolia; 15) Monticeddu; 16) P.le Laridda; 17) Giampilieri; 18) P.le
Dinareddi; 19) Pezzolo; 20) Briga Superiore; 21) San Paolo; 23) Upper Schiavo river basin; 24) S. Stefano; 25) P.le Furcu; 26) and Villa Parotta. These areas were selected for the presence
of clusters of PS characterized by the highest ground deformation velocities. (See online version for color image).
112 A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
(DRPC), the analysis of the available SAR data highlighted areas
affected by deformation located also outside of the landslide perimeter,
suggesting that the latter was slightly underestimated (Fig. 10).
The integration between space-borne and ground-based data was
useful to analyze the ground deformation related to the San Fratello
landslide in different time intervals (from 1992 to 2013) and with
different LOS and ground resolutions. The new landslide boundary
contains areas affected by deformation, which were detected by the
C-band sensors during the pre landslide event phase, and by both the
COSMO-SkyMed satellite and the GBInSAR system during the postevent
phase. The new landslide area increased in extension from 1 to
1.2 km2
, and was validated also through field surveys aimed at the
buildings and infrastructures damage assessment.
The accuracy obtained in landslide mapping and monitoring in
correspondence of a large inhabited area affected by slope instability,
such as the town of San Fratello, proved to be crucial in risk management
activities and in the preparation of emergency plans.
4.3.5. Residual risk zonation (Altolia)
Immediately after the 1st of October event, a microzoning of landslide
risk was performed for all the populated centers affected by landslide
phenomena. The purposes of the microzoning were (i) to prioritize
risk mitigation works and (ii) to identify areas characterized by low to
null risk. As an example, the case of the zonation of Altolia is presented
and discussed. This case is particularly relevant because it includes both
the residual risk related to occurrence of fast moving debris flows and
a slow moving pre-existing slide. The populated center is classified
according to four zones with different risk levels (Table 2). Most of
buildings and houses close to the two riverbeds (Giampilieri and its
right-bank tributary the Mandarina, Fig. 10) were classified with different
level of risk (from 1 to 3) when several mud, earth and debris flows
occurred. These phenomena were triggered as a consequence of the
high water content of the soil located on the slopes of the Mandarina
creek. The mobilized material moved as a channeled flow within the
creek, damaging the buildings on the left and right banks, and finally
flowing through a culvert, into the Giampilieri creek. Following the
ranking displayed in Table 2, these buildings were classified in the
highest risk level both for their location (inside or close to the riverbed)
and for their damage degree. The southeast part of the village was
classified as medium to high risk (level 2) even though it was not
affected by debris- or mud-flow phenomena. This level of risk was due
to the presence of a slow-moving pre-existing slide, whose presence
was inferred from the analysis of radar data, photo interpretation and
field survey. The central part of the village, located on the hilltop, far
from the riverbeds, is characterized by negligible geomorphological
and hydrological risk. Furthermore PS data highlight that this area is
not affected by slow-moving landslide phenomena (Fig. 10).
5. Discussion
Each product presented in this paper for landslide territorial
management shows both advantages and limitations, which can be
enhanced or reduced by choosing the appropriate scale of application,
geomorphological/geological background or typology of phenomenon.
Fig. 7. The Altolia (Messina) hotspot. (a) ERS dataset; (b) ENVISAT dataset; and (c) geomorphological map where the landforms were divided into pre-existing and formed after the 1st of
October event. PS datasets clearly suggest that the biggest landslide located in the southeastern sector is active at least since 1992. (See online version for color images).
113A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
For example all the products based on PSInSAR data (ground deformation
velocity maps, hotspot mapping, VSlope projection, updated LIM,
building deformation velocity map and partially the risk zonation
mapping) were affected by the limitations of this technique. The most
significant limits of PSInSAR data mainly refer to the different technical
parameters of the satellite systems used, which result in unlike acquisition
features and consequent results (Ferretti et al., 2001). The presently
available SAR data, from both historical archives and currently operational
missions, do not allow the measurement of deformations faster
than few tens of cm per year (Hanssen, 2005; Crosetto et al., 2010).
Velocities compromising the PS processing depend on the employed
SAR wavelengths and on the satellite repeat cycles; these recorded
displacement rates resulted: (i) from about 15 cm/yr up to 20 cm/yr
in C-band (for ERS/ENVISAT and RADARSAT satellites, respectively);
(ii) about 25 cm/yr for TerraSAR-X; (iii) 68 cm/yr for COSMO-SkyMed
system with 4 operating satellites in X band. Consequently, rapid
surface movements characterized by higher motion rates could not be
correctly detected and analyzed. Another limit affecting the multifrequency
PS data is the possibility to improve detection and monitoring
of hydrogeological events not in real-time, but only mainly in the
Fig. 8. Examples of building deformation velocity maps: (a) building deformation velocity map by using the ENVISAT dataset; (b) building deformation velocity map by using the CSK
dataset; and (c) damage assessment map of San Fratello. The red line in (a) and (b) represents the 2010 landslide boundary. The black lines in (b) represent the main scarps developed
after the 2010 landslide. It is worth noting that most damaged buildings (c) correspond to buildings affected by the highest deformation velocity measured by CSK data (b) suggesting that
after almost 2 years from the main event the area is still subjected to residual deformations.
Fig. 9. Integration between space borne (CSK) and GBInSAR measurement. Comparison between the DRPC landslide boundary, obtained through field investigation, and the new boundary
inferred by the integration between PS and GBInSAR data. On the background: the GBInSAR cumulative displacement map (from March 2010 to March 2013) and the COSMO- SkyMed
data (from Bardi et al., 2014). (See online version for color image).
114 A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
deferred time, and, partially, in the near- real time. By contrast, this
approach allows the analysis of the temporal evolution of ground deformation
up to 1992 (in case of ERS-1/2 satellite imagery availability).
Concerning the PS spatial sampling, vegetation causes decorrelation in
SAR images, while stable scatterers coincide with rocky outcrops and
man-made built-up structures (Werner et al., 2003). Thus, while in
some areas, like urbanized zones, PS interferometry obtains a reasonably
good sampling, in other areas (vegetated, rural and forested)
benchmarks are very reduced and PS interferometry tends to systematically
fail. This is particularly true when using higher frequencies
(e.g. C-band), whereas lower frequencies (e.g. L-band sensors) are less
affected by atmospheric absorption, less attenuated by the vegetation
canopy, providing a higher number of PS.
The PSInSAR technique allows the measurement of slow and very
slow ground deformation (up to few millimeters per year) representing
a useful support technique for landslide hazard mapping and monitoring
activities, detecting movements that can occur several weeks or months
before a catastrophic failure (Bardi et al., 2014). This technique can be
used both at the regional or local scale.
The limitation related to the PSInSAR technique can be solved or
reduced by integrating its results with those obtained through other
techniques and available data. For example, at the local scale, the
ground deformation velocity can be assessed more frequently in time
using a ground based interferometer which can acquire on routine
displacement data. A geodatabase makes easier the integration among
different techniques and data, often represented by geographic
information (thematic maps, field measurements, remote sensing
measurements).
The availability of a geodatabase integrating data and products
strictly devoted to landslide hazard represents a great support for the
public authorities in charge of landslide risk management (Pagliara
et al., 2014). The opportunity to obtain a framework as complete as
possible and continuously updated with new products and data
organized in a geodatabase facilitates the utilization of all the available
information in each phase of the emergency cycle, both in real and in
deferred time.
The geodatabase structure and its workflow to produce each product
were shared and discussed with public authorities and users at different
administrative levels (national, regional and local), which were potentially
involved in landslide risk management activities. This close collaboration
among scientists, stakeholders and users was useful to make the
scientists and researchers able to provide products and disseminate
these with a geodatabase, meeting the user needs, both in terms of
value added of the products and of their accessibility. The availability
of all the data used to generate the basic and advanced products
supports the users to understand the specific product value and their
usefulness of application. In addition, the validation of the products
and the structure of the geodatabase and the pre-operational application
were useful to integrate the results in the operative workflow and
in the most useful ways. The challenge to continuously update the
geodatabase with new products and to increase the usability of its
structure will allow satisfying new user requirements. This will make
Fig. 10. Landslide risk zonation for the Altolia village (Messina). The four risk levels were decided following the description of Table 2 considering buildings characteristics (damages and
distance from the riverbed) and the detected ground deformation velocity by SAR sensors. Light blue arrows correspond with the main flow directions of the Giampilieri and Mandarina
creeks.
115A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
this work useful and suitable to be developed and built in other test sites
in order to aim at the geodatabase and product availability with the
same quality at national level.
In Table 3 the landslide product integration performed in the
Civil Protection emergency cycle is outlined, particularly regarding the
technical assessment process implemented by the close collaboration
of research institutes and public authorities. The technical assessments
obtained were inputs for the user technical activities.
Products obtained through the geodatabase of the Messina Province
are currently used by the authorities during the different Civil
Protection phases. For example, a combined use among the updated
landslide inventory map, the ground deformation velocity maps and
the susceptibility map is useful for prevention and preparedness
actions. These products highlight the most unstable areas and the
degree to which an area can be affected by future slope failure at the
regional scale. These information allow to decide the priority for the
definition of the Civil Protection emergency plan for the most hazardous
areas.
Local scale products (e.g. residual risk zonation, hotspot mapping,
updated landslide inventory map) are more useful during the emergency
phase to plan the preliminary stabilization interventions, an early
warning system and the evacuation of the population. These products
were successfully used for the Giampilieri and Altolia emergency
and evacuation plan. The management of the recovery phase needs
products useful to monitor the landslide area and to plan preliminary
interventions. During this phase products such as residual risk zonation,
local scale deformation analysis (using the GBInSAR or field measurements)
and the building deformation velocity maps are of fundamental
importance. These products and activities were used to monitor the San
Fratello landslide and to make people and workers' activities safe.
The products generated after a landslide emergency event will
become the starting point for the next prevention and preparedness
phase.
6. Conclusions
Ground deformations, and in particular landslides, represent one of
the most hazardous phenomena that cause economic and human losses.
For these reasons Civil Protection and local authorities are strongly
interested in assessing the analysis of the spatial distribution of landslide
hazard and risk, for planning and decision-making activities, and
in particular for a proper management of the different Civil Protection
cycle phases (preparedness/prevention, emergency and recovery).
Updated landslide inventory maps provide useful information on
location, occurrence, landslide-type, area involved, state of activity,
failure mechanisms, triggering factors, and damage assessment. These
tasks require a very large amount of information, such as the detection
and mapping of unstable areas, thematic and environmental data, and
element at risk evaluation.
Detection and mapping of unstable areas are essential procedures
that can be produced and updated by handling different data in a
dedicated geodatabase where available geographic information and
new acquired remote sensing data (earth observation, ground based,
aerial) can be profitably exploited to improve the quality of landslide
inventory maps, landslide assessment and susceptibility.
In this work, a geodatabase of the Messina Province is presented and
described highlighting the capability of the GIS environment to obtain
useful products for the management of the landslide-induced hazard.
The geodatabase contains available ancillary data and new acquired
SAR and optical data, which were used to produce regional and local
maps useful for the Civil Protection Department to manage the 2009
and 2010 events post emergency phases. For example the building
deformation velocity maps produced in San Fratello and the risk zonation
realized in Altolia were used to plan recovery intervention and
urban development. The geodatabase was also used to produce ground
deformation velocity maps and the susceptibility map of the Messina
Province, which can be used to detect the most unstable areas for the
planning of mitigation activities. The structure of the database allows
the updating of some of the products presented in cases of acquisition
of new data, thus providing continuously updated maps.
The presented database was designed following the requirements
of the local and national Civil Protection authorities which today constantly
use it to manage the landslide hazard of the Messina Province
at regional and local levels. This work proves that a very complete
geodatabase is a flexible tool suitable to contribute to the management
of the landslide hazard and risk.
Acknowledgments
The research leading to these results received funding from the
European Union Seventh Framework Programme (FP7/2007–2013)
under grant agreement No. 242212 (Project DORIS). This study was
founded by the Seventh Framework Programme of European Commission:
Landslide Modelling and Tools for vulnerability assessment preparedness
and recovery management (LAMPRE). Federico Di Traglia is
supported by a post-doc fellowship founded by the Regione Toscana
(UNIPI-FSE) under the project RADSAFE (UNIPI-4) in the framework
of the research agreement between DST-UNIPI, DST-UNIFI and Ellegi–
Lisalab. The risk zonation for the village of Altolia has been performed
in collaboration with the National Civil Protection and the Department
of Civil and Environmental Engineering of the University of Catania.
We thank two anonymous reviewers and the editor for their suggestions
which greatly improved this work.
Table 3
Applications of landslide products in the Civil Protection emergency cycle.
Civil Protection phase Landslide products Technical assessment User technical activities
Prevention and
preparedness
- Landslide inventory map
- Ground-deformation velocity map
- Susceptibility map
- Definition of landslide hazard scenarios - Definition of the hydrogeological management
plan
- Definition of Civil Protection emergency plan
Emergency - Event and pre-existing Landslide
inventory map
- Residual risk zonation
- ground deformation velocity map
- Hot spot mapping
- Residual risk evaluation
- Displacement monitoring
- Evaluation the magnitude of the landslide event
- Plan of the preliminary stabilization interventions
- Early warning
- Evacuation
- Relocation
- Adoption and/or updating of civil protection
emergency plan
- Adoption of the preliminary stabilization plan
Recovery - Ground-deformation velocity maps
- Landslide inventory maps
- Residual risk zonation
- Susceptibility map
- Projected ground deformation
velocity map
- Local scale deformation analysis
- Building deformation velocity map
- Evaluation of the preliminary stabilization
interventions
- Plan of the final stabilization interventions
- Updating of landslide hazard scenarios
- Evaluation of local geomorphological hazard for
relocated settlements
- Updating (if necessary) of the preliminary
stabilization plan
- Adoption of the final stabilization plan
- Updating of the hydrogeological management
plan
- Updating of the Civil Protection emergency
plans
116 A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
References
Ardizzone, F., Basile, G., Cardinali, M., Casagli, N., Del Conte, S., Del Ventisette, C., Fiorucci,
F., Garfagnoli, F., Gigli, G., Guzzetti, F., Iovine, G., Mondini, A.C., Moretti, S., Panebianco,
M., Raspini, F., Reichembach, P., Rossi, M., Tanteri, L., Terranova, O., 2012. Landslide
inventory map for the Briga and the Giampilieri catchments, NE Sicily, Italy. J. Maps
8, 176–180.
Bardi, F., Frodella, W., Ciampalini, A., Bianchini, S., Del Ventisette, C., Gigli, G., Fanti, R.,
Moretti, S., Basile, G., Casagli, N., 2014. Integration between ground based and satellite
SAR data in landslide mapping: the San Fratello case study. Geomorphology 223,
45–60.
Battistini, A., Segoni, S., Manzo, G., Catani, F., Casagli, N., 2013. Web data mining for
automatic inventory of geohazards at national scale. Appl. Geogr. 43, 147–158.
Bellotti, F., Bianchi, M., Colombo, D., Ferretti, A., Tamburini, A., 2014. Advanced InSAR
Techniques to Support Landslide Monitoring. In: Pardo-Igúzquiza, E., GuardiolaAlbert,
C., Heredia, J., Moreno-Merino, L., Durán, J.J., Vargas-Guzmán, J.A. (Eds.),
Mathematics of Planet Earth, Lecture Notes in Earth System Sciences. Springer, Berlin
Heidelberg, pp. 287–290.
Bianchini, S., Herrera, G., Notti, D., Garcìa-Moreno, I., Mora, O., Moretti, S., 2013. Landslide
activity maps generation by means of Persistent Scatterer Interferometry. Remote
Sens. 5, 6198–6222.
Bianchini, S., Ciampalini, A., Raspini, F., Bardi, F., Di Traglia, F., Moretti, S., Casagli, N., 2014a.
Multi-temporal evaluation of landslide movements and impacts on buildings in San
Fratello (Italy) by means of C-band and X-band PSI data. Pure Appl. Geophys.
http://dx.doi.org/10.1007/s00024-014-0839-2.
Bianchini, S., Tapete, D., Ciampalini, A., Di Traglia, F., Del Ventisette, C., Moretti, S., Casagli,
N., 2014b. Multi-temporal evaluation of landslide-induced movements and damage
assessment in San Fratello (Italy) by means of C-band and X-band PSI data. In:
Pardo-Igúzquiza, E., Guardiola-Albert, C., Heredia, J., Moreno-Merino, L., Durán, J.J.,
Vargas-Guzmán, J.A. (Eds.), Mathematics of Planet Earth, Lecture Notes in Earth
System Sciences. Springer, Berlin Heidelberg, pp. 257–261.
Billi, A., Funiciello, R., Minelli, L., Faccenna, C., Neri, G., Orecchio, B., Presti, D., 2008. On the
cause of the 1908 Messina tsunami, southern Italy. Geophys. Res. Lett. 35, L06301.
http://dx.doi.org/10.1029/2008GL033251.
Breiman, L., 2001. Random forests. Mach. Learn. 45, 5–32.
Carbone, S., Catalano, S., Grasso, M., Lentini, F., Monaco, C., 1990. Carta Geologica della
Sicilia centro-orientale. Università di Catania, scale 1:50,000.
Casagli, N., Tibaldi, A., Merri, A., Del Ventisette, C., Apuani, T., Guerri, L., Fortuny-Guasch, J.,
Tarchi, D., 2009. Deformation of Stromboli Volcano (Italy) during the 2007 eruption
revealed by radar interferometry, numerical modelling and structural geological
field data. J. Volcanol. Geotherm. Res. 182, 182–200.
Casagli, N., Catani, F., Del Ventisette, C., Luzi, G., 2010. Monitoring, prediction, and early
warning using ground-based radar interferometry. Landslides 7, 291–301.
Cascini, L., Fornaro, G., Peduto, D., 2009. Analysis at medium scale of low-resolution
DInSAR data in slow-moving landslide-affected areas. ISPRS J. Photogramm. Remote
Sens. 64, 598–611.
Cascini, L., Fornaro, G., Peduto, D., 2010. Advanced low- and full-resolution DInSAR map
generation for slow-moving landslide analysis at different scales. Eng. Geol. 112,
29–42.
Catani, F., Lagomarsino, D., Segoni, S., Tofani, V., 2013. Exploring model sensitivity issues
across different scales in landslide susceptibility. Nat. Hazards Earth Syst. Sci. 13,
2815–2831.
Ciampalini, A., Cigna, F., Del Ventisette, C., Moretti, S., Liguori, V., Casagli, N., 2012.
Integrated geomorphological mapping in the north-western sector of Agrigento
(Italy). J. Maps 8, 136–145.
Ciampalini, A., Bardi, F., Bianchini, S., Frodella, W., Del Ventisette, C., Moretti, S., Casagli, N.,
2014. Analysis of building deformation in landslide area using multisensor PSInSAR™
technique. Int. J. Appl. Earth Obs. Geoinf. 33, 166–180.
Colombo, A., Lanterni, L., Ramasco, M., Troisi, C., 2005. Synthetic GIS-based landslide
inventory as the first step for effective landslide-hazard-management. Landslides 2,
291–301.
Corrado, S., Aldega, L., Balestrieri, M.L., Maniscalco, R., Grasso, M., 2009. Structural
evolution of the sedimentary accretionary wedge of the alpine system in Eastern
Sicily: thermal and thermochronological constraints. Geol. Soc. Am. Bull. 121,
1475–1490.
Crosetto, M., Monserrat, O., Iglesias, R., Crippa, B., 2010. Persistent Scatterer Interferometry:
potential, limits and initial C-band and X-band comparison. Photogramm. Eng. Remote
Sens. 76, 1061–1069.
Cruden, D.M., Varnes, D.J., 1996. Landslide types and processes. In: Turner, A.K., Shuster,
R.L. (Eds.), Landslides: Investigation and Mitigation. Spec. Rep 247. Transp. Res.
Board, Washington D.C., pp. 36–75.
De Guidi, G., Lanzafame, G., Palano, M., Puglisi, G., Scaltrito, A., Scarfì, L., 2013. Multidisciplinary
study of the Tindari Fault (Sicily, Italy) separating ongoing contractional and
extensional compartments along the active Africa–Eurasia convergent boundary.
Tectonophysics 588, 1–17.
Del Ventisette, C., Garfagnoli, F., Ciampalini, A., Battistini, A., Gigli, G., Moretti, S., Casagli,
N., 2012. An integrated approach to the study of catastrophic debris-flow: geological
hazard and human influence. Nat. Hazards Earth Syst. Sci. 12, 2907–2922.
Del Ventisette, C., Ciampalini, A., Manunta, M., Calò, F., Paglia, L., Ardizzone, F., Mondini,
A.C., Reichembach, P., Mateos, R.M., Bianchini, S., Garcia, I., Füsi, B., Deak, Z.V., Radi,
K., Graniczny, M., Kowalski, Z., Piatkowska, A., Przylucka, M., Retzo, H., Strozzi, T.,
Colombo, D., Mora, O., Sanches, F., Herrera, G., Moretti, S., Casagli, S., Guzzetti, F.,
2013. Exploitation of large archives of ERS and ENVISAT C-band SAR data to
characterize ground deformations. Remote Sens. 5, 3896–3917.
Dellow, G.D., Glassey, P.J., Lukovic, B., Word, P.R., Morrison, B., 2003. Data sources of the
New Zealand landslide database. Geophys. Res. Abstr. 5, 13867.
Devoli, G., Strauch, W., Chávez, G., Høeg, K., 2007. A landslide database for Nicaragua: a
tool for landslide-hazard management. Landslides 4, 163–176.
Dewey, J.F., Helman, M.L., Turco, E., Hutton, D.H.W., Knott, S.D., 1989. Kinematics of the
western Mediterranean. In: Coward, M.P., Dietrich, D., Parks, R.G. (Eds.), Alpine
tecnonics. Spec. Publ. Geol. Soc. London 45, pp. 265–283.
Di Paolo, L., Olivetti, V., Corrado, S., Aldega, L., Balestrieri, M.L., Maniscalco, R., 2014.
Detecting the stepwise propagation of the Eastern Sicily thrust belt (Italy): insight
from thermal and thermochronological constraints. Terra Nova 26, 363–371.
Di Stefano, E., Agate, M., Incarbona, A., Russo, F., Sprovieri, R., Bonomo, S., 2012. Late
Quaternary high uplift rates in northeastern Sicily: evidences from calcareous
nannofossils and benthic and planktonic foraminifera. Facies 58, 1–15.
Di Traglia, F., Intrieri, E., Nolesini, T., Bardi, F., Del Ventisette, C., Ferrigno, F., Frangioni, S.,
Frodella, W., Gigli, G., Lotti, A., Tacconi Stefanelli, C., Tanteri, L., Leva, D., Casagli, N.,
2014a. The ground-based InSAR monitoring system at Stromboli volcano: linking
changes in displacement rate and intensity of persistent volcanic activity. Bull.
Volcanol. 76, 1–18.
Di Traglia, F., Cauchie, L., Casagli, N., Saccorotti, G., 2014b. Decrypting geophysical signals
at Stromboli Volcano (Italy): integration of seismic and ground‐based InSAR displacement
data. Geophys. Res. Lett. 41, 2753–2761.
Ferretti, A., Prati, C., Rocca, F., 2000. Non-linear subsidence rate estimation using
permanent scatterers in differential SAR interferometry. IEEE Trans. Geosci. Remote
Sens. 38, 2202–2212.
Ferretti, A., Prati, C., Rocca, F., 2001. Permanent scatterers in SAR interferometry. IEEE
Trans. Geosci. Remote Sens. 39, 8–20.
Ferretti, A., Fumagalli, A., Novali, F., Prati, C., Rocca, F., Rucci, A., 2011. A new algorithm for
processing interferometric data-stacks: SqueeSAR. IEEE Trans. Geosci. Remote Sens.
49, 3460–3470.
Gaspar, J.L., Coulart, C., Queiroz, G., Silveira, D., Gomes, A., 2004. Dynamic structure and
data sets of a GIS database for geological risk analysis in the Azores volcanic islands.
Nat. Hazards Earth Syst. Sci. 4, 233–242.
Gigli, G., Fanti, R., Canuti, P., Casagli, N., 2011. Integration of advanced monitoring and
numerical modeling techniques for the complete risk scenario analysis of rockslides:
the case of Mt. Beni (Florence, Italy). Eng. Geol. 120, 48–59.
Gustavvson, M., Seijmonsbergen, A.C., Kolstrup, E., 2008. Structure and contents of a new
geomorphological GIS database linked to a geomorphological map — With an
example from Linden, central Sweden. Geomorphology 95, 335–349.
Guzzetti, F., 2000. Landslide fatalities and evaluation of landslide risk in Italy. Eng. Geol.
58, 89–107.
Guzzetti, F., Cardinali, M., Reichembach, P., 1994. The AVI project: a bibliographical and
archive inventory of landslides and floods in Italy. Environ. Manag. 18, 623–633.
Guzzetti, F., Mondini, A.C., Cardinali, M., Fiorucci, F., Santangelo, M., Chang, K.T., 2012.
Landslide inventory maps: new tools for an old problem. Earth Sci. Rev. 112, 42–66.
Hanssen, R.F., 2005. Satellite radar interferometry for deformation monitoring: a prior
assessment of feasability and accuracy. Int. J. Appl. Earth Obs. Geoinf. 6, 253–260.
Herrera, G., Gutiérrez, F., García-Davalillo, J.C., Guerrero, J., Notti, D., Galve, J.P., FernándezMerodo,
J.A., Cooksley, G., 2013. Multi-sensor advanced DInSAR monitoring of very
slow landslides: the Tena Valley case study (Central Spanish Pyrenees). Remote
Sens. Environ. 128, 31–43.
Intrieri, E., Di Traglia, F., Del Ventisette, C., Gigli, G., Mugnai, F., Luzi, G., Casagli, N., 2013.
Flank instability of Stromboli volcano (Aeolian Islands, Southern Italy): integration
of GB-InSAR and geomorphological observations. Geomorphology 201, 60–69.
Klose, M., Damm, B., Terhorst, B., 2014. Landslide cost modeling for transportation
infrastructures: a methodological approach. Landslides http://dx.doi.org/10.1007/
s10346-014-0481-1.
Lagomarsino, D., Segoni, S., Fanti, R., Catani, F., Casagli, N., 2014. Regional scale landslide
susceptibility mapping in Emilia Romagna (Italy) as a tool for early warning.
Landslide Science for a Safer Geoenvironment, pp. 443–449.
Meisina, C., Notti, D., Zucca, F., Ceriani, M., Colombo, A., Poggi, F., Roccati, A., Zaccone, A.,
2013. The use of PSInSAR™ and SqueeSAR™ techniques for updating landslide inventories.
In: Margottini, C., Canuti, P., Sassa, K. (Eds.), Landslide Science and Practice.
Springer, Berlin Heidelberg, pp. 81–87.
Messina, A., Somma, R., Careri, G., Carbone, G., Macaone, E., 2004. Peloritani continental
crust composition (southern Italy): geological and petrochemical evidences. Boll.
Soc. Geol. Ital. 123, 405–441.
Nolesini, T., Di Traglia, F., Del Ventisette, C., Moretti, S., Casagli, N., 2013. Deformations and
slope instability on Stromboli volcano: integration of GBInSAR data and analog
modeling. Geomorphology 180, 242–254.
Notti, D., Davalillo, J.C., Herrera, G., Mora, O., 2010. Assessment of the performance of
X-band satellite radar data for landslide mapping and monitoring: Upper Tena Valley
case study. Nat. Hazards Earth Syst. Sci. 10, 1865–1875.
Notti, D., Herrera, G., Bianchini, S., Meisina, C., García-Davalillo, J.C., Zucca, F., 2014. A methodology
for improving landslide PSI data analysis. Int. J. Remote Sens. 35, 2186–2214.
Pagliara, P., Basile, G., Cara, P., Corazza, A., Duro, A., Manfrè, B., Onori, R., Proietti, C.,
Sansone, V., 2014. Integration of earth observation and ground-based HR data in
the civil protection emergency cycle: the case of the DORIS project. In: PardoIgúzquiza,
E., Guardiola-Albert, C., Heredia, J., Moreno-Merino, L., Durán, J.J., VargasGuzmán,
J.A. (Eds.), Mathematics of Planet Earth, Lecture Notes in Earth System
Sciences. Springer, Berlin Heidelberg, pp. 263–266.
PAI-Piano Stralcio di Bacino per l'Assetto Idrogeologico, 2012. AdB Regione Sicilia.
http://www.sitr.regione.sicilia.it/pai.
Pereira, S., Zêzere, J.L., Quaresma, I.D., Bateira, C., 2014. Landslide incidence in the North
of Portugal: analysis of a historical landslide database based on press releases and
technical reports. Geomorphology 214, 514–525.
Plank, S., Singer, J., Minet, C., Thuro, K., 2012. Pre-survey suitability evaluation of the
differential synthetic aperture radar interferometry method for landslide monitoring.
Int. J. Remote Sens. 33, 6623–6637.
117A. Ciampalini et al. / Geomorphology 249 (2015) 103–118
Raspini, F., Moretti, S., Casagli, N., 2013. Landslide mapping using SqueeSAR data:
Giampilieri (Italy) case study. In: Margottini, C., Canuti, P., Sassa, K. (Eds.), Landslide
Science and Practice. Springer, Berlin Heidelberg, pp. 147–154.
Raucoules, D., Bourgine, B., Michele, M., Le Gozannet, G., Closset, L., Bremmer, C.,
Veldkamp, H., Tragheim, D., Bateson, L., Crosetto, M., Agudo, M., Engdahl, M., 2009.
Validation and intercomparison of persistent scatterers interferometry: PSIC4 project
results. J. Appl. Geophys. 68, 335–347.
Righini, G., Pancioli, V., Casagli, N., 2012. Updating landslide inventory maps using
Persistent Scatterer Interferometry (PSI). Int. J. Remote Sens. 33, 2068–2096.
Suprechi, L., Floris, M., Ghirotti, M., Genevois, R., Jaboyedoff, M., Stead, D., 2010. Technical
note: implementation of a geo-database of published and unpublished data on the
catastrophic Vaiont landslide. Nat. Hazards Earth Syst. Sci. 10, 865–873.
Tarchi, D., Casagli, N., Moretti, S., Leva, D., Sieber, A.J., 2003a. Monitoring landslide
displacements by using ground-based radar interferometry: application to the
Ruinon landslide in the Italian Alps. J. Geophys. Res. 108, 101–114.
Tarchi, D., Casagli, N., Fanti, R., Leva, D., Luzi, G., Pasuto, A., Pieraccini, M., Silvano, S., 2003b.
Landslide monitoring by using ground-based SAR interferometry: an example of
application to the Tessina landslide in Italy. Eng. Geol. 68, 15–30.
Thomas, M.F.H., Bodin, S., Redfern, J., Irving, D.H.B., 2010. A constrained African craton
source for the Cenozoic Numidian Flysch: implications for the palaeogeography of
the western Mediterranean basin. Earth-Sci. Rev. 101, 1–23.
Van Den Eeckhaut, M., Hervás, J., 2012. State of the art of national landslide databases
in Europe and their potential for assessing landslide susceptibility, hazard and risk.
Geomorphology 139–140, 545–558.
van Westen, C.J., Van Asch, T.W.J., Soeters, R., 2005. Landslide hazard and risk zonation;
why is it still so difficult? Bull. Eng. Geol. Environ. 65, 167–184.
Werner, C., Wegmuller, U., Strozzi, T., Wiesmann, A., 2003. Interferometric Point Target
Analysis for deformation mapping. Proceedings of IGARSS 2003, Toulouse (Francia)
7, pp. 4362–4364.
Zêzere, J.L., Pereira, S., Tavares, A.O., Bateira, C., Trigo, R.M., Quaresma, I., Santos, P.P.,
Santos, M., Verde, J., 2014. DISASTER: a GIS database on hydro-geomorphologic
disaster in Portugal. Nat. Hazards 72, 503–532.
118 A. Ciampalini et al. / Geomorphology 249 (2015) 103–118