See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/225951175

Quality Assessment of the Italian Landslide Inventory Using GIS Processing

Article in Landslides · December 2010

DOI: 10.1007/s10346-010-0213-0

CITATIONS READS

240 712

3 authors:

Alessandro Trigila Carla Iadanza

Institute for Environmental Protection and Research (ISPRA) Institute for Environmental Protection and Research (ISPRA)
57 PUBLICATIONS 1,354 CITATIONS 48 PUBLICATIONS 1,434 CITATIONS

SEE PROFILE SEE PROFILE

Daniele Spizzichino
Institute for Environmental Protection and Research (ISPRA)
191 PUBLICATIONS 1,495 CITATIONS

SEE PROFILE

All content following this page was uploaded by Carla Iadanza on 05 February 2018.

The user has requested enhancement of the downloaded file.

 Author's personal copy
Original Paper

Landslides (2010) 7:455–470 Alessandro Trigila I Carla Iadanza I Daniele Spizzichino

DOI 10.1007/s10346-010-0213-0
Received: 28 July 2009
Accepted: 9 April 2010 Quality assessment of the Italian Landslide Inventory
Published online: 25 June 2010
© Springer-Verlag 2010
using GIS processing

Abstract Landslides constitute one of the most important hazard models at a national level and for land use planning
natural hazards in Italy as they are widespread and result in and management.
considerable damage and fatalities every year. The Italian
Landslide Inventory (IFFI) Project was launched in 1999 with The Italian Landslide Inventory
the aim of identifying and mapping landslides over the entire The Italian Landslide Inventory (IFFI Project), with funding of
Italian territory. The inventory currently holds over 480,000 4.1 million euro from the Italian Government, is conducted by
landslides and has been available by means of Web services Institute for Environmental Protection and Research (ISPRA)-
since 2005. The aim of this study is to define quality indices Geological Survey of Italy and by the 21 Regions and Self-
for evaluation of the homogeneity and completeness of the Governing Provinces. ISPRA has the task of organizing and
IFFI database. In order to estimate the completeness of the coordinating the activities, producing guidelines, and disseminat-
landslide attribute information, a heuristic approach has been ing information via the Internet, while the regions and the self-
used to assign weighting values to significant parameters governing provinces collect historical documents, archive data,
selected from the landslide data sheet. The completeness and and map areas affected by landslides.
homogeneity of the landslide mapping has been evaluated by
means of three different methods: an area-frequency distribu-
tion analysis; the proximity of the landslides surveyed to urban Data collection and representation
areas; variation of the landslide index within the same The methodology applied to collect, harmonize, integrate, and
lithology. The quality indices have allowed identification of validate the landslide data is based on aerial photo interpretation,
areas with a high level of completeness and critical areas in historical documents, and field surveys. The classification of the
which the data collected have been underestimated or are not landslides is based on schemes by Varnes (1978), Cruden and
very accurate. The quality assessment of collected and stored Varnes (1996), and recommendations by the International
data is essential in order to use the IFFI database for definition Association of Engineering Geology (1990), the International
and implementation of landslide susceptibility models and for Geotechnical Societies UNESCO Working Party on Word Land-
land use planning and management. slide Inventory (1990, 1991, 1993a, b, 1994), and the International
Union of Geological Science Working Group on Landslides (1995).
Keywords Landslide . Inventory . Quality index . The IFFI integrates information from (a) national projects as
Natural hazards . GIS . Italy AVI-Inventory of information on sites historically affected by
landslides and floods in Italy for the period 1918–2000 (Guzzetti
Introduction et al. 1994), SCAI-Special project for the study of unstable towns,
The exposure of the Italian territory to natural hazards is CARG-Geological map of Italy, scale 1:50,000; (b) landslide
remarkably high. After earthquakes, landslides cause the greatest inventories by Regions, River Basin Authorities, research insti-
losses in terms of victims and damage to buildings, infrastructure, tutes, and universities; (c) river basin plans (PAI - Law 267/98);
and environmental and cultural heritage. Expansion of urban (d) Civil Defense Emergency Declarations (Law 225/92); (e)
areas and infrastructure over the last 50 years has led to a marked national and local public archives; and (f) scientific papers and
increase in landslide risk. In recent times, disastrous landslide technical reports.
events have occurred in Piemonte in 1994, in Versilia in 1996, in The information related to each landslide has been collected
Sarno and Quindici-Campania Region in 1998, in a large sector of using a landslide data sheet (Fig. 1) organized in three informa-
northern Italy in 2000, in Val Canale-Friuli Venezia Giulia Region tion levels: the first level contains basic data on landslide location,
in 2003, and in Messina in 2009 (ANPA-ARPAT 1998; ARPA type of movement and state of activity; the second level provides
Piemonte 2003; Calcaterra et al. 2000; Del Prete et al. 1998). data on geometry, geologic units, lithology, geotechnical proper-
The Italian Landslide Inventory (IFFI) Project was launched ties, land use, causes, and date of occurrence; the third level gives
in 1999 with the aim of identifying and mapping landslides detailed information on damage, investigation process, and
throughout Italy on the basis of standardized criteria. The remedial measures for risk reduction (Amanti et al. 2001).
inventory currently holds over 480,000 landslides covering an The IFFI geo-database contains geographic features of land-
area of approximately 20,500 km2, which is equivalent to 6.8% of slides (Fig. 2), an alphanumeric archive of attributes and a
the Italian territory. multimedia archive (photos, videos, documents in PDF format).
The aim of this paper is to apply methods for analyzing the Landslides are represented by a geo-referenced point (located at
quality of landslide databases and to specifically evaluate the highest point of the crown), by a polygon if the landslide may
the quality of the IFFI. This evaluation is essential for use of be mapped at the adopted survey scale or by a line when the
the database, both for the development of susceptibility and width is very narrow, as in the case of debris flows. The mapping

Landslides 7 & (2010) 455

 Author's personal copy
Original Paper

Fig. 1 Section of the IFFI landslide data sheet

scale is 1:10,000 for most of Italy. However, a smaller scale area/total area × 100) calculated on a 1×1-km grid (Trigila
(1:25,000) has been used in mountainous and sparsely populated 2007).
areas. The landslide ID is a code which uniquely identifies each
landslide within the whole Italian territory. Web services for the dissemination of information via the internet
The Italian Landslide Inventory currently holds 482,272 The IFFI Inventory was published on the Internet in 2005
landslides. Figure 3 shows the Landslide Index (landslide (http://www.sinanet.apat.it/progettoiffi) with the aim of dissem-

456 Landslides 7 & (2010)

 Author's personal copy

Fig. 2 IFFI cartographic representation

inating landslide information to central and local public The WMS service of the IFFI has been developed according
administrations, research institutes, geologists, engineers, and to European Directive INSPIRE-Infrastructure for spatial informa-
citizens. Two services have been designed: the WebGIS tion in Europe-2007/2/CE (The European Parliament and the
application Cart@net-IFFI and the WMS service. Cart@net-IFFI Council of the European Union 2007) and the Open GIS
allows geographical data to be explored and detailed informa- Consortium specifications (Open Geospatial Consortium 2006).
tion to be obtained on the most important landslide param- This interoperable service enables the user to build a customized
eters. The application manages large amounts of spatial data map overlaying the IFFI landslides with other layers stored on the
with extremely rapid response times and integrates the vector user’s own computer or provided by different Internet map
datasets, served by ArcSDE and ArcIMS Server (ESRI) and servers. The user can access the WMS service through a WMS
Oracle DB, with the imagery served by the Image Web Server viewer, such as the one provided by the INSPIRE Geo-portal, or
(ER Mapper) in ECW (Enhanced Compressed Wavelet) format. through a Desktop GIS (e.g. ESRI ArcGIS®).
The user can view the landslides of the IFFI together with
other vector layers (e.g., urban areas—Corine Land Cover Quality assessment of the Italian landslide database
2000, roads and railways, administrative boundaries, and
drainage network) and raster layers (digital terrain model, Quality of landslide inventories
digital orthophoto TerraItaly it2000, Landsat satellite images, The quality of a landslide inventory depends mainly on the
and IGM topographic map; Fig. 4). method used to create the inventory. For example, the use

Landslides 7 & (2010) 457

 Author's personal copy
Original Paper

Fig. 3 Landslide Index (%) calculated on a 1×1-km grid

merely of aerial photo interpretation does not allow the photos (Carrara et al. 1995; Guzzetti 2002; Wills and Mc Crink
landslide date of occurrence to be determined and makes it 2002). In order to set up a more complete inventory, it is
difficult to identify dormant landslides which have been necessary to use a multi-temporal series of aerial photos
significantly remodeled by farming activities, covered by thick which enable reconstruction of the developmental dynamics of
vegetation, or are located in urban areas. The correct the landslides, in addition to allowing a better identification
identification and mapping of the landslides depends not and classification. With regard to the use of historical and
only on the degree of conservation of the morphological archive data, the greatest limitations are found in the
evidence (slopes, tensile cracks, etc.) but also on the imprecise description of the landslide location and the non-
experience of the photo interpreter and the scale of the aerial homogeneous spatial distribution of the landslides due to the

458 Landslides 7 & (2010)

 Author's personal copy

Fig. 4 IFFI Project WebGIS interface

underestimation of events in sparsely populated areas and, With regard to the quality of the IFFI, while the data
therefore, with a lower number of manmade structures were being checked, all geometrical features and attribute
(Guzzetti et al. 1999). In addition, while there is a great deal information passed a series of controls before being stored in
of information on the damage, data on the type of landslide the geo-database. In particular, two sets of quality control
and its dynamics are often lacking (Carrara et al. 2003). The tasks were performed: the first set identified omissions,
integration of the historical data and the photo interpretation duplicate values in the ID field, digital mapping anomalies,
analyses enables the abovementioned limitations to be over- geo-positional errors, etc. The second set inspected the IFFI
come and to significantly increase the quality of the with other datasets: AVI (Inventory of information on sites
inventory. historically affected by landslides and floods in Italy for the
To date, there are no well-established scientific criteria in period 1918–2000), SCAI (Special project for the study of
the literature for evaluating the accuracy, reliability, and unstable towns), CARG (Geological map of Italy, scale
completeness of a landslide inventory. This evaluation is 1:50,000), and PAI (River basin plan).
carried out in some studies by comparing inventories This study has carried out a further evaluation of the
prepared for the same area (Ardizzone et al. 2002; Carrara completeness of the information contained in the alphanu-
et al. 1992; Galli et al. 2008), while in other studies, the meric database, and the homogeneity and level of detail of
frequency-area statistics are determined and analyzed. the landslide mapping, as described below.

Landslides 7 & (2010) 459

 Author's personal copy
Original Paper
Table 1 Most significant parameters of the IFFI Landslide data sheet
Parameters Level of IFFI data sheet Landslides (No.) Landslides (%) Weight
Type of movement 1 455,566 94.46 6
State of activity 1 423,086 87.73 2
Date of most recent observation 1 403,886 83.75 2
Method used to determine type of movement and state of activity
Not determined 1 6,998 1.45 0
1 method 386,393 80.12 1.5
2 methods 85,019 17.63 3
3 or 4 methods 3,485 0.72 6
Instrumental monitoring 377 0.08 12
Damages 1 73,691 15.28 1
Aerial photo interpretation metadata 1 165,470 34.31 1
Archives 1 100,392 20.82 1
National legislations 1 62,903 13.04 0.5
Civil Defense Emergency Declarations 1 422 0.09 0.5
Bibliography 1 37,957 7.87 2
Lithology 2 53,805 11.16 3
Geotechnical properties 2 49,393 10.24 2
Rock mass structure 2 12,539 2.60 0.5
Joint spacing 2 3,575 0.74 0.5
Bedding attitude 2 3,304 0.69 0.5
Weathering 2 5,090 1.06 0.5
Depth of surface of rupture Dr (m) 2 2,951 0.61 3
Land cover 2 59,385 12.31 1
Slope aspect 2 68,612 14.23 1
Rate of movement 2 8,806 1.83 1
Material 2 35,184 7.30 0.5
Distribution of activity 2 6,705 1.39 1
Style of activity 2 6,211 1.29 0.5
Causes 2 23,485 4.87 3
Damages 2 19,908 4.13 2
Damages (water course) 2 1,929 0.40 2
Date of occurrence 2 4,767 0.99 1
Dating of most significant event
Certain data 2 3,364 0.70 3
Uncertain data—day 2,690 0.56 1.5
Uncertain data—month 4,021 0.83 1
Uncertain data—year 2,569 0.53 0.5
Superficial water 3 11,482 2.38 1
Springs 3 6,409 1.33 1
Groundwater 3 2,487 0.52 1
Precursory signs 3 4,250 0.88 3
Technical reports 3 4,960 1.03 2

460 Landslides 7 & (2010)

 Author's personal copy

Table 1 (continued)
Parameters Level of IFFI data sheet Landslides (No.) Landslides (%) Weight
Investigations and monitoring 3 1,523 0.32 3
Remedial measures 3 5,278 1.09 3
Photos – 5,841 1.21 3
Attached documents – 522 0.11 10

Completeness of the IFFI landslides’ attribute information

In order to estimate the completeness of the attribute informa-
tion, each landslide has been assigned a rating. A heuristic
approach has been used to assign weighting values to significant
parameters selected from the landslide data sheet and to the
adopted investigation method (Table 1).
Thirty-seven parameters were selected from the landslide
data sheet belonging to the 1st, 2nd, and 3rd levels. A weight
of 6 was attributed to the type of movement which is the
fundamental parameter for understanding and describing the
landslide phenomena. A value of 3 was assigned to 8
Fig. 5 Most used investigation methods parameters considered particularly important (e.g. lithology,

Table 2 Completeness rating of landslide attribute information

Landslides (No.) Low (%) Medium (%) High (%) Very high (%)
Italia 482,272 10 75 10 5
Piemonte 35,023 29 65 5 1
Valle d'Aosta 4,359 89 10 0 0
Lombardia 130,538 1 96 3 1
Bolzano-Bozen 1,995 0 2 14 84
Trento 9,385 9 27 35 29
Veneto 9,476 14 36 40 10
Friuli Venezia Giulia 5,253 3 66 18 13
Liguria 7,515 22 76 0 2
Emilia Romagna 70,037 0 99 0 1
Toscana 39,517 54 35 10 2
Umbria 34,544 6 70 10 13
Marche 42,522 13 86 1 0
Lazio 10,548 0 32 62 6
Abruzzo 8,493 2 37 54 7
Molise 23,940 0 96 2 2
Campania 23,430 3 89 1 6
Puglia 843 0 1 36 63
Basilicata 9,187 0 3 81 16
Calabria 9,417 0 30 40 30
Sicilia 4,727 5 24 38 34
Sardegna 1,523 0 0 1 99

Landslides 7 & (2010) 461

 Author's personal copy
Original Paper

Fig. 6 Relationship between the rating of completeness Very high and the number of landslides for each Italian Region

depth of the sliding surface, date of occurence, cause,

consolidation works). A lower weight of between 2 and 0.5
was attributed to the other parameters, with the exception of
the value of 10 assigned in the presence of technical reports
attached to the data sheet. These reports often contain
detailed information on the stratigraphy, laboratory tests,
geological reports, etc. With regard to the method used, a
higher rating has been assigned as the number of methods
applied for identification and classification of each landslide
has increased—with a maximum value of 12 for landslides
with instrumental monitoring.
With regard to the percentage of the data sheet fields
which have been completed, the type of movement and the
state of activity have been filled in, respectively, for 94.5% and
87.7% of the landslides surveyed. A much lower percentage of
the 2nd and 3rd level fields have been completed, such as, for
example the “cause” parameter with 4.9%. With regard to the
method used, 80% of the landslides of the IFFI Inventory
have been identified using only one of the planned methods
(Fig. 5). In order to improve the accuracy and the reliability
of the inventory, the information is currently being updated
and integrated with more than one investigation method. In Fig. 7 Area-frequency distribution of the IFFI

462 Landslides 7 & (2010)

 Author's personal copy

order to evaluate the completeness of the alphanumeric previously existing inventories but also to the greater or lesser
archive of the IFFI, four classes of values have been defined: degree of use of the aerial photo interpretation, field surveys, or
Low for values <10; Medium 10–15; High 15–21; Very high ≥ 21 historical and archives data in the methodology adopted by the
(Table 2). individual Regions/Self-Governing Provinces. The data relative to
It has been seen from the analysis carried out at a three Regions of Southern Italy (Basilicata, Calabria and Sicily)
national scale that 75% of the landslides have a Medium value are underestimated compared with the actual instability situation
with respect to the completeness of the alphanumeric data- since, to date, the landslide survey activities been concentrated
base. At a regional scale, it has been seen that the quality mainly in urban areas or areas with main road and rail
depends on the number of landslides which populate each infrastructures.
regional database, such as, for example, in the case of The completeness and homogeneity of the landslide
Sardinia (Fig. 6). mapping has been evaluated in this study by means of three
different analysis methods: (a) area-frequency statistics; (b)
Completeness and homogeneity of the IFFI landslide mapping analysis of the proximity of the landslides surveyed to urban
There is a good level of detail in the landslide mapping for the areas; (c) analysis of the variation of the landslide index within
IFFI Project: over a total of 482,272 landslides surveyed, 377,016 the same lithology.
(78%) are represented by a polygon, 50,651 (11%) by a line and
only 53,236 (11%) by a point. If the distribution of the landslides Frequency-area statistics
throughout Italy is analyzed (Fig. 3), it is seen that there is a The use of this method is based on the assumption that landslides,
certain lack of homogeneity of the landslide data within the just like other natural hazards (e.g., earthquakes, forest fires), show
Alpine Arc due not only to the different levels of detail of the fractal or scale invariant size statistics, i.e., the cumulative

Table 3 Landslide area statistics

Landslides Landslides represented Landslides represented Maximum Minimum Most frequent
(No.) by a polygon (No.) by a polygon (%) area (km2) area (m2) area (m2)
Italia 482,272 377,027 78.18 40.92 100 2,990
Piemonte 35,023 23,820 68.01 21.77 1,601 2,254
Valle d'Aosta 4,359 1,212 27.80 25.56 2,066 11,295
Lombardia 130,538 75,104 57.53 34.67 100 2,837
Bolzano-Bozen 1,995 759 38.05 40.92 101 113
Trento 9,385 5,195 55.35 16.09 823 1,790
Veneto 9,476 4,380 46.22 4.01 110 2,254
Friuli-Venezia 5,253 4,987 94.94 7.22 111 566
Giulia
Liguria 7,515 7,482 99.56 4.84 220 8,972
Emilia Romagna 70,037 62,641 89.44 4.27 1,600 1,790
Toscana 39,517 37,397 94.64 10.26 114 5,661
Umbria 34,544 31,511 91.22 2.87 100 1,790
Marche 42,522 39,787 93.57 4.05 1,606 5,661
Lazio 10,548 8,796 83.39 2.81 385 11,295
Abruzzo 8,493 8,064 94.95 9.37 3,344 22,536
Molise 23,940 23,484 98.10 3.25 360 5,661
Campania 23,430 18,097 77.24 7.43 410 14,219
Puglia 843 411 48.75 2.55 10,039 11,295
Basilicata 9,187 9,152 99.62 4.45 406 1,790
Calabria 9,417 9,324 99.01 11.75 324 5,661
Sicilia 4,727 4,360 92.24 14.86 163 179
Sardegna 1,523 1,053 69.14 21.20 219 1,129

Landslides 7 & (2010) 463

 Author's personal copy
Original Paper
distribution of the frequency of the landslide area may be described limit of the size of the landslide in a certain geological and
over intervals with varying degrees of magnitude by a power-law geomorphological context, while it is seen by others as the
distribution (Hergarten 2002; Malamud and Turcotte 1999; Reich- effect of the sub-sampling of the small-sized landslides due
enbach et al. 2002). This relationship has been found to be valid, to the mapping scale and the mapping resolution (Guthrie
regardless of the types of movement present, the size of the and Evans 2004; Malamud et al. 2004; Stark and Hovius
inventory or the triggering mechanisms (e.g., short and intense or 2001). The sub-sampling may be due to the objective
prolonged rainfall, earthquakes). impossibility of observing the landslides due to an insuffi-
From an analytical point of view, the cumulative frequency cient resolution of the aerial photos or the presence of
distribution is expressed by (1): shadow areas in the aerial photos, tree cover, or cancellation
of the morphological evidence of the landslides. With regard
NCL ¼ CAa ð1Þ to this latter cause, it should be noted that some types of
L
small landslides are frequently remodeled by farming activ-
ities and may only be identified for a short time after the
where NCL is the cumulative number of landslides with an area date of occurrence of the landslide. Large landslides, on the
greater than or equal to the area AL. In a diagram with logarithmic other hand, leave traces which may be seen for hundreds of
axes, the Eq. 1 is represented by a straight line, with α equal to the years.
gradient of the interpolated straight line and C is equal to the
intercept on the y axis.
Similarly, the noncumulative frequency distribution is given by
(2): Table 4 Mapping coverage in the areas far away from urban settlements

% Landslides
dNCL Region/ Quality
(distance > 750 m
 ¼ C0 Ab
L ð2Þ Self-Governing Province Index
dAL from urban areas)
Piemonte 44.5
with b ¼ a þ 1 (α exponent of the cumulative distribution). Valle d'Aosta 59.9
In order to obtain the noncumulative area-frequency distribu- Lombardia 66.4
tion starting from the cumulative distribution, use may be made of
Bolzano-Bozen 49.2
the derivatives method, which calculates the derivative of NCL(A) as
Trento 54.5
the gradient of the cumulative curve which interpolates a
determined number of points (generally 5) around a particular Veneto 32.6
value of A (Frattini et al. 2003). The area-frequency distribution Friuli-Venezia Giulia 63.7
may also be estimated by using the logarithmic binning method Liguria 32.3
(Guzzetti et al. 2003; Hovius et al. 1997, 2000). The data sample is Emilia Romagna 51.0
subdivided into intervals (bins) with constant width along a Toscana 52.9
logarithmic axis, that is, keeping the difference logsi–logsi−1 Umbria 59.6
constant, where si and si−1 are the ends of the ith interval). Since Marche 51.5
the bin widths grow with increasing landslide size, the logarithmic Lazio 39.1
binning method overcomes the problem that large landslides are
Abruzzo 43.2
generally very few.
Molise 71.6
The empirical frequency is calculated as the ratio between
the number of landslides which fall within each interval and Campania 47.4
the width of the interval itself. A diagram is prepared in the Puglia 62.2
bilogarithmic plane of si snii1 versus bsi , where ni is the number Basilicata 25.9
of landslides which fall within the ith interval and bsi is the Calabria 12.6
average value of the interval. In the case of the power law, the Sicilia 54.7
noncumulative distribution obtained by means of the loga- Sardegna 61.0
rithmic binning method is equivalent to a cumulative distri-
bution, so the gradient of the straight line which interpolates
the data in the bilogarithmic plane is equivalent to the
exponent α of the cumulative distribution. Quality Index
The area-frequency distribution is described by the High Landslide(%) > Mean + St. Dev.
power law of a certain interval. Below a certain area value, Medium Mean - St.Dev. < Landslide(%) < Mean + St.Dev.
there is a drop in frequency, known as rollover (deviation
Low Landslide(%) < Mean - St. Dev.
from power law scaling). The rollover which may be seen in
all the frequency curves shown in the literature is inter-
preted by some authors as real, that is, due to the physical Mean = 49; Standard Deviation = 14,5

464 Landslides 7 & (2010)

 Author's personal copy

The area-frequency distribution for the 377,016 landslides Self-Governing Provinces (min. area, max. area and most
of the IFFI represented by a polygon has been determined frequent area; Table 3) enabled the adopted work scale and
by means of the logarithmic binning method. The data the level of detail of the mapping resolution to be evaluated.
sample was classified in intervals with a constant width of The most complete inventories at a regional level correspond
log si  log si1 ¼ 0:05. to values less than the most frequent landslide area.
The area-frequency distribution follows a power-law, with
the straight line interpolated in the bilogarithmic plane having Proximity to urban areas
a gradient of 2.07, which is equivalent to an exponent 3.07 of Landslide inventories are more accurate in areas close to
the noncumulative distribution (Fig. 7). The exponent urban centers or linear communication infrastructures since
obtained has an excellent correspondence with the values in the historical information linked to the damage caused by
the literature. Studies carried out on numerous inventories for landslides is more numerous, more detailed studies are
landslides throughout the world, in a multitude of climatic carried out as part of the territorial plans (e.g. River Basin
and physiographic regions, have shown a range of values for Plans - PAI) and access to the sites is easier for performance
the exponent b ¼ 2:5  0:5 (Guzzetti et al. 2002). of the ground surveys. The spatial distribution of the land-
The power-law scaling is valid for four area magnitudes, slides in the IFFI Inventory was, therefore, evaluated on the
from a maximum area of 40 km2 to a minimum area of 1× basis of the distance from the urban areas. For this purpose a
10–2 km2, which corresponds to the rollover. The most 750-m wide buffer was generated around the Italian urban
frequent area in the inventory is 2,990 m2. areas extracted from the Prior 10 k DB (Intesa GIS Project -
At a national scale, the inventory may be deemed to be scale 1:10,000) and the percentage of landslides which fall
complete for landslides with areas greater than 10,000 m2. The outside the buffer was calculated. Three quality classes were
analysis of the area-frequency distributions for the 21 Regions/ then defined on the basis of the mean and standard deviation

Fig. 8 Landslide mapping in Molise Region: the figure shows a full coverage of the territory: urban areas (in red); landslides far from urban areas (in blue); landslides
inside the 750 m buffer (in cyan)

Landslides 7 & (2010) 465

 Author's personal copy
Original Paper

Fig. 9 Lithological map of Italy

of the 21 regional values (Table 4). It may be seen from the Landslide density and lithology
analyses that four regions have a Low quality index, which is Lastly, the spatial completeness of the IFFI was evaluated by
indicative of a non-uniform territorial coverage of the analyzing the variation of the landslide index within the
investigations, 14 Regions have a Medium index, and three same outcropping lithology. The Lithological Map 1:500,000
have a High index, as in the case of the Molise Region (Fig. 9), derived from the Geological Map of Italy by
(Fig. 8). Geological Survey of Italy (scale 1:500,000-1983), was used as

466 Landslides 7 & (2010)

 Author's personal copy

input by grouping the 128 geological units into 13 lithological include territories at an elevation >600 m (Trigila and
classes. The analysis was carried out within the mountainous- Iadanza 2008).
hilly territory (Fig. 10) which is the one in which landslides Figure 11 shows the landslide index calculated for the lithological
may potentially occur. This territory was obtained by using a class “Clays and clays & marls” in the mountainous-hilly territory.
20×20-m DEM. The alluvial plains include territories with The particularly low landslide index values for the “Clays and clays &
an elevation <300 m and slope <3°; the hills have a slope >3° marls” in Basilicata, Calabria and Sicily are indicative of an
or an elevation between 300 and 600 m; the mountains underestimate of landslides in these areas (Fig. 12).

Fig. 10 Landform classification derived from DEM

Landslides 7 & (2010) 467

 Author's personal copy
Original Paper

Fig. 11 Landslide Index (%) related to clays, clays, and marls in mountainous-hilly territory

Conclusions level parameters of the landslide data sheet were low. At a

An evaluation of the quality of the IFFI Inventory was regional level, the completeness of the alphanumeric data
carried out by means of four types of analyses, both depends on the number of landslides held in each regional
statistical and spatial in a GIS environment. The first database. The other analyses, aimed at evaluating the
analysis, aimed at evaluating the completeness of the completeness of the mapping database, included a study of
alphanumeric database, gave an “average” value for the the area-frequency distribution, the distance of the surveyed
quality at a national level, highlighting that the “movement landslides from the urban centers and the landslide index on
type” field was completed for 94.5% of the landslides, whilst a lithological basis. The area-frequency analysis showed that
the percentages of the completed fields for the 2nd and 3rd the distribution follows a power-law relationship with an

468 Landslides 7 & (2010)

 Author's personal copy

Fig. 12 Landslide Index % (calculated over a 250-m wide grid) related to clays, clays, and marls in Basilicata Region

exponent which has a good correspondence with the values susceptibility models. The analyses have also detected critical
in the literature. The frequency distribution differs from the areas in which the data collected have been underestimated
power-law (rollover) at areas equal to 10,000 m2. With regard or are not very accurate and for which it will be necessary to
to the analysis of the proximity of the landslide surveys to integrate the data in the future. However, the aims of the
the urban centers, it was seen that the territorial coverage of IFFI Project may be considered fully achieved bearing in
the investigations is almost uniform for the majority of the mind that in 1999, before the start of the Project, approx-
Regions. Comparison of the landslide indices for areas with imately 70,000 landslides had been recorded by the regions
the same outcropping lithology has enabled identification of and self-governing provinces.
critical areas in which the data collected has been under- The IFFI, in terms of the number of landslides surveyed, the
estimated, as in the case for the Basilicata Region. full coverage of the Italian territory, and the detailed landslide
The quality indices have allowed identification of repre- mapping, represents a capable tool for hazard assessment and
sentative areas in terms of homogeneity and reliability of the land use planning. The IFFI Project methodology could be
IFFI Project data, which may be used as training and test adopted at the European level for the realization of a unique
areas for the implementation of quantitative landslide and homogeneous inventory of landslides in Europe.

Landslides 7 & (2010) 469

 Author's personal copy
Original Paper
References IAEG–International Association Engineering Geology Commission on Landslides (1990)
Suggested Nomenclature for Landslides. Bull Int Assoc Eng Geol 41:13–16
Amanti M, Bertolini G, Chiessi V, De Nardo MT, Ramasco M (2001) Guida alla IUGS/WGL–International Union of Geological Science Working Group on Landslides
compilazione della scheda frane IFFI. Geological Survey of Italy, Roma. http://www. (1995) A suggested method for describing the rate of movement of a landslide. Bull
mais.sinanet.apat.it/cartanetiffi/doc/Allegati_Tecnici/Allegato_1.pdf. Accessed 9 Int Assoc Eng Geol 52:75–78
March 2010 Malamud BD, Turcotte DL (1999) Self-organized criticality applied to natural hazards.
ANPA-ARPAT (1998) 19 giugno 1996: alluvione in Versilia e Garfagnana. Un caso di Natural Hazards 20:93–116
studio. ARPAT, Firenze, 315 pp Malamud BD, Turcotte DL, Guzzetti F, Reichenbach P (2004) Landslide inventories and
Ardizzone F, Cardinali M, Carrara A, Guzzetti F, Reichenbach P (2002) Impact of mapping their statistical properties. Earth Surface Processes and Landforms 29(6):687–711
errors on the reliability of landslide hazard maps. Nat Hazards Earth Syst Sci 2:3–14 Open Geospatial Consortium (2006) OpenGIS® Web Map Server Implementation
ARPA Piemonte (2003) Eventi alluvionali in Piemonte 13–16 ottobre 2000. ARPA Specification (Version: 1.3.0). http://www.opengeospatial.org/standards/wms.
Piemonte, Torino, 327 pp Accessed 9 March 2010
Calcaterra D, Parise M, Palma B, Pelella L (2000) Multiple debris flow in volcanoclastic Reichenbach P, Guzzetti F, Malamud BD, Turcotte DL (2002) Comparison of two
materials mantling carbonate slope. In: Proc. 2nd Int. Conf. on Debris-Flow Hazard landslide triggering events using frequency-area statistics. In: Deidda R, Mugnai A,
Mitigation, Taipei, Taiwan 16–18 August 2000. Balkema, Rotterdam, pp 99–107 Siccardi F (eds) Proceedings of the 3rd EGS Plinius Conference on Mediterranean
Carrara A, Cardinali M, Guzzetti F (1992) Uncertainty in assessing landslide hazard and Storms, Italy, 1–3 October 2001, pp 403–408
risk. ITC Journal, The Netherlands 2:172–183 Stark CP, Hovius N (2001) The characterization of landslide size distributions.
Carrara A, Cardinali M, Guzzetti F, Reichenbach P (1995) GIS technology in mapping Geophysical Research Letters 28(6):1091–1094
landslide hazard. In: Carrara A, Guzzetti F (eds) Geographical information systems in The European Parliament and the Council of the European Union (2007) Directive 2007/2/EC
assessing natural hazards. Kluwer, Dordrecht, pp 135–175 establishing an Infrastructure for Spatial Information in the European Community
Carrara A, Crosta G, Frattini P (2003) Geomorphological and historical data in assessing (INSPIRE), UE, Brussels. http://inspire.jrc.ec.europa.eu. Accessed 9 March 2010
landslide hazard. Earth Surf Proc Land 28:1125–1142 Trigila A (ed.) (2007) Rapporto sulle frane in Italia, Il Progetto IFFI – Metodologia,
Cruden DM, Varnes DJ (1996) Landslide types and processes. In: Turner AK, Schuster RL risultati e rapporti regionali (Rapporti 78/2007). APAT, Roma, 681 pp. http://www.
(eds) Landslides investigation and mitigation (Special report 247). Transportation apat.gov.it/site/it-IT/APAT/Pubblicazioni/Rapporti/Documento/rapporto_2007_78.
Research Board, Washington, D.C., pp pp 36–75 html. Accessed 9 March 2010
Del Prete M, Guadagno FM, Hawkins AB (1998) Preliminary report on the landslides of 5 Trigila A, Iadanza C (2008) Landslide in Italy – Special Report 2008 (Rapporti 83/2008).
May 1998 in Campania, Southern Italy. Bull. Eng. Geol. Env. 57:113–129 ISPRA, Roma, 32 pp. http://www.apat.gov.it/site/it-IT/APAT/Pubblicazioni/Rapporti/
Frattini P, Crosta GB, Ceriani M, Fossati D (2003) Inventario delle frane e dei dissesti Documento/rapporto_83_08_landslides.html. Accessed 9 March 2010
della Regione Lombardia: analisi statistica e probabilistica per una valutazione Varnes DJ (1978) Slope movements types and processes. In: Schuster RL, Krizek RJ (eds)
preliminare della pericolosità. Atti del 1° Convegno Nazionale AIGA, pp 427–448 Landslides: analysis and control (Special Report 176). Transportation Research Board,
Galli M, Ardizzone F, Cardinali M, Guzzetti F, Reichenbach P (2008) Comparing landslide National Academy of Science, Washington, pp pp 11–33
inventory maps. Geomorphology 94:268–289 Wills CJ, Mc Crink TP (2002) Comparing landslide inventories: the map depends on the
Guthrie RH, Evans SG (2004) Magnitude and frequency of landslides triggered by a method. Environ Eng Geosci 8(4):279–293
storm event, Loughborough Inlet, British Columbia. Nat. Hazards Earth Syst. Sci. WP/WLI-International Geotechnical Societies UNESCO Working Party on World Landslide
4:475–483 Inventory (1990) A suggested method for reporting a landslide. Bull Int Assoc Eng
Guzzetti F (2002) Landslide Cartography, Hazard Assessment and Risk Evaluation: Geol 41:5–12
Overview, Limits and Prospective. Proceedings of 3rd MITCH Workshop Floods, WP/WLI-International Geotechnical Societies UNESCO Working Party on World Landslide
droughts and landslides who plans, who pays, Potsdam (Germany), November 24– Inventory (1991) A suggested method for a landslide summary. Bull Int Assoc Eng
26, 2002 Geol 43:101–110
Guzzetti F, Cardinali M, Reichenbach P (1994) The AVI Project: a bibliographical and WP/WLI-International Geotechnical Societies UNESCO Working Party on World Landslide
archive inventory of landslides and floods in Italy. Environmental Management Inventory (1993) A suggested method for describing the activity of a landslide. Bull
18:623–633 Int Assoc Eng Geol 47:53–57
Guzzetti F, Carrara A, Cardinali M, Reichenbach P (1999) Landslide hazard evaluation: a WP/WLI-International Geotechnical Societies UNESCO Working Party on World Landslide
review of current techniques and their application in a multi-scale study, central Inventory (1993) Multilingual Landslide Glossary. Bitech Publisher, Richmond, British
Italy. Geomorphology 31:181–216 Columbia, Canada, 59 pp
Guzzetti F, Malamud BD, Turcotte DL, Reichenbach P (2002) Power-law correlations WP/WLI-International Geotechnical Societies UNESCO Working Party on World Landslide
of landslide areas in central Italy. Earth and Planetary Science Letters 195:169– Inventory (1994) A suggested method for reporting landslide causes. Bull Int Assoc
183 Eng Geol 50:71–74
Guzzetti F, Reichenbach P, Wieczorek GF (2003) Rockfall hazard and risk assessment in
the Yosemite Valley, California, USA. Nat. Hazards Earth Syst. Sci. 3:491–503
Hergarten S (2002) Self-organized criticality in earth systems. Springer, Berlin, p 272 pp
Hovius N, Stark CP, Allen PA (1997) Sediment flux from a mountain belt derived by A. Trigila ()) : C. Iadanza : D. Spizzichino
landslide mapping. Geology 25:231–234 ISPRA (Institute for Environmental Protection and Research),
Hovius N, Stark CP, Hao-Tsu C, Jiun-Chuan L (2000) Supply and Removal of via Vitaliano Brancati 48, 00144, Roma, Italy
Sediment in a Landslide Dominated Mountain Belt: Central Range, Taiwan. J Geol e-mail: [email protected]
108:73–89 URL: http://www.sinanet.apat.it/progettoiffi

470 Landslides 7 & (2010)

View publication stats