Quantitative vulnerability assessment of buildings to debris

International Conference
Analysis and Management of Changing Risks for Natural Hazards
18-19 November 2014 l Padua, Italy
Quantitative vulnerability assessment of buildings to debris-flows in Fella River Basin
using run-out modeling and damage data from the 29th of August 2003 event
R. Ciurean1, L. Chen2, H. Hussin2, C. van Westen2, T. Glade1, S. Frigerio3, A. Pasuto3
Department of Geography and Regional Research, University of Vienna, Vienna, Austria
Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente,
Enschede, Netherlands
CNR-IRPI, National Research Institute for Geo-hydrological Protection, Padova, Italy
Corresponding author details:
R. Ciurean, Department of Geography and Regional Research, University of Vienna,
Universitätstrasse 7, 1010, Vienna, Austria, E-mail: [email protected]
Vulnerability, empirical curves, debris flow, run-out modeling, buildings, Fella River, Italy
Extreme rainfall events trigger floods and debris flows with considerable consequences in
mountain regions worldwide. The reduction of possible human and material losses is
dependent on the assessment of risks as well as design and implementation of effective
reduction strategies. These in turn rely not only on the analysis of the magnitude and
frequency of the harmful events but also on the comprehensive evaluation of exposed
elements and their vulnerability. Vulnerability is defined as the degree of loss to a given
element or set of elements at risk resulting from the occurrence of a hazard of a given
magnitude in a given area, expressed as a percentage of loss (between 0: no damage, to 1:
total damage) (Varnes, 1984). Although, conceptual frameworks for quantitative vulnerability
estimations have been developed (Papathoma-Köhle et al., 2011), including more recent
studies on uncertainty analysis (Eidsvig et al., 2014, Totschnig and Fuchs, 2013, Kaynia et
al., 2008), there is still a great need for advancing methodologies in areas where they are
essential for risk mitigation investments and improved decision-making. This study aims to
develop a set of vulnerability curves for buildings impacted by debris flows in the Fella River
Basin (Eastern Italian Alps). The area experienced a major rainfall event in August 2003
which triggered more than a thousand debris flows resulting in significant economic losses
and two casualties. Two methodological approaches are applied for the estimation of
process intensities: one based on dynamic run-out modeling (regional scale), and another
based on interpretation of aerial and photographic documentation (local scale). The results
are compared with damage values estimated shortly after the event, as well as vulnerability
curves from the literature. Asset characteristics are collected through field work and desktop
mapping, and population distribution and value for each building is calculated based on
census and real estate data. The obtained vulnerability curves together with population
distribution and building monetary value are subsequently used in an exposure and risk
analysis at regional scale in Fella River Basin.
The study site (247 km2) pertains to the Fella River Basin (Canal del Ferro and Val Canale,
Eastern Italian Alps), a region environmentally (i.e. structural, lithological, morphological,
climatic) prone to hazards such as debris flows and floods, flash-floods as well as
earthquakes. The area presents high national interest due to its strategic position which
resulted in time in urbanization and development of dense infrastructure networks (European
International Conference
Analysis and Management of Changing Risks for Natural Hazards
18-19 November 2014 l Padua, Italy
communication and energy corridor) (Malek et al., 2014). Nevertheless, it has a low
population density (between 2,26/km2 in Dogna and 22/km2 in Tarvisio, ISTAT 2014) with
fluctuations during tourism activities – the main source of income for the inhabitants. The
administrative units overlapping the study area are: Malborghetto-Valbruna, Pontebba,
Tarvisio, and Dogna communes (Fig. 1).
Figure 1. (Left) The Fella River Basin Study area; (right) damages caused by the 2003 event in the
study area (©Civil Protection, FVG)
The latest major hydro-meteorological event in the Fella area occurred in August 2003 (Fig.
1), when approximately 1 million cubic meter of debris flow deposits and river flooding
produced damages to the infrastructure and buildings on an area of about 765 km 2, resulting
in two casualties (Civil Protection, 2012). Reports of the Civil Protection show that the
economic losses caused by this event in the Malborghetto-Valbruna, Pontebba, Tarvisio, and
Dogna communes reach a total of 389 million Euros, out of which 49% was registered in
Malborghetto Valbruna commune.
The methodological steps leading to the development of the vulnerability curves for both
regional and local assessment approaches are presented in Figure 2. Two general sets of
data were used, one leading to the estimation of process intensities (debris flow data) and
the other, leading to the estimation of assets‟ degree of loss (buildings data).
Debris flow data
For the first approach (A, in Figure 2), the debris flow inventories were produced by the
Italian Landslide AVI (CNR-IRPI, 2014) and IFFI Projects (ISPRA, 2014), Geological Survey
of Friuli-Venezia Giulia Region (FVG), and landslide experts from the University of Trieste.
The inventory contains 273 debris flow source area points and run-out polygons located
mostly along the Val Canale slopes. Rainfall data (1976 – 2011) was used to determine three
return periods by extreme value distribution analysis: 100 – 500 years for „major‟ events (e.g.
August 2003), 25 – 100 years for moderate‟ events (e.g. pre-2003 event), 10 – 25 years for
minor‟ events (2003 - 2011 events), 1 – 10 years for „frequent‟ events. Two Digital Elevation
Models (2003, 2007) with a pixel size of 10 m were obtained from the Civil Protection of the
FVG Region and utilized for the susceptibility and run-out analysis. The methodology
developed by Hussin et al. (2013) was applied to generate a debris flow susceptibility map
for the Fella River area. This map together with the event inventory map for different return
International Conference
Analysis and Management of Changing Risks for Natural Hazards
18-19 November 2014 l Padua, Italy
periods were used to identify the release areas necessary for the run-out modeling. In
addition to the high susceptibility class cells, the planar curvature and slope angle thresholds
were employed to better define the source areas. The debris flow run-out modeling was
performed using Flow-R (Horton et al., 2013), a software developed at the University of
Lausanne, which uses a distributed empirical model for regional susceptibility assessments
of debris flows. The model requires the minimum travel angle and the maximum velocity for
each return period; these parameters were estimated based on back calibration of a limited
number of events for each return period. Once the source areas maps for each return period
were produced and integrated into the run-out model, the kinetic energy and maximum runout probability were obtained.
Figure 2. Methodological steps used for vulnerability assessment of buildings to debris-flows in Fella
River Basin (A - regional scale: using dynamic run-out modeling; B – local scale: using photodocumentation)
In order to estimate the intensity value for the modelled runouts, the probability values were
transferred into impact pressures using linear transfer functions. These functions are based
on two factors: (1) the spatial distribution and variation of the probability values within the
debris flow morphology (from the debris flow channels and transportation zones to the end of
the deposit zones at the debris fans) and (2) the estimated impact pressures in the field
based on damage assessments of past events.
In comparison with the first, in the second approach (B, in Figure 2) debris flow intensities
were not modelled but estimated using a methodology developed by (Papathoma-Köhle et
al., 2012). Information regarding the intensity was acquired from the interpretation of highresolution oblique air photos and on-site photographic documentation of the resulting
consequences collected by the Civil Protection of the FVG Region immediately after the
August 2003 event. The height of the maximum debris and water flow was measured from
the indicated marks on the building walls in relation with the building height. Following the
International Conference
Analysis and Management of Changing Risks for Natural Hazards
18-19 November 2014 l Padua, Italy
intensity assessment, the recorded damages were analyzed and compared with
compensation costs provided by the local authorities of Malborghetto Valbruna municipality
for each affected building. The damage to the interior assets was not taken into account.
After evaluating the impact on buildings and calculating the buildings value (see “Buildings
data” section) the degree of loss was expressed as the percentage loss from the total
building value. Each building was represented as a point in a XY coordinate system (x axis
representing the intensity, y axis -, the degree of loss). Non-linear regression approaches
can be applied to obtain best fitting functions. Cumulative distributions fulfil the mathematical
requirements to model the vulnerability curves (define the depending variable i.e. degree of
loss in a both sided confined interval [0, 1]; they are steady and monotonic increasing with
the interval of its explaining variable i.e. intensity) (Totschnig et al., 2011).
Buildings data
An initial digital building dataset was provided by Civil Protection of the Friuli-Venezia Giulia
region. The dataset contained information about the location (X, Y coordinates), geometry
(height, area, volume, base and top elevation) and a general land use classification. As this
information was not sufficient for vulnerability and risk assessment, firstly a survey was
carried out using Google Street View for updating the building locations, and characterizing
the building occupancy classes, building materials types and number of floors. Moreover,
many of the buildings that were destroyed in 2003, or demolished later, had to be corrected,
as well as the buildings that were constructed later had to be inserted. Subsequently, a
fieldwork was carried out for validating the desktop mapping results using a mobile GIS
application (Figure 3).
Figure 3. Update and validation of building inventory using Google Street View and field work
Additionally, population and building value information was generated based on existent data
from the Italian National Institute of Statistics (ISTAT, http://www.istat.it/) and the Italian
Revenue Agency (Agenzia delle Entrate, http://www.agenziaentrate.gov.it) as follows:
a) Population values available for the administrative units were subdivided over the
buildings in the area, using a dasymetric mapping approach, taking into account the
International Conference
Analysis and Management of Changing Risks for Natural Hazards
18-19 November 2014 l Padua, Italy
occupancy class (focusing mainly on residential buildings) and building size. Two
different scenarios were taken into account for population modeling: one normal situation
where most of the population is located in the residential areas, and a tourist season
scenario, where also additional population is distributed over hotels and other tourist
b) Minimum and maximum market values in Euro per Square meter (for the second
semester of 2013) are given for different buildings with major land use types. These were
further subdivided based on the zone where the buildings are located according to the
Real Estate Observatory data (Osservatorio del Mercato Immobiliare, Agenzia Entrate –
OMI) (center, periphery, rural zones). Based on the building footprint area, and the
number of floors, the total floor space area was calculated, which was then multiplied by
the minimum and maximum values, so that the result is the estimated minimum and
maximum building value.
The population distribution and building value in the studied area was subsequently used to
calculate the vulnerability, exposure and risk to debris flows and floods.
Regional scale vulnerability assessment using dynamic run-out modeling
Vulnerability curves for debris flow impact pressure and debris flow height were generated
for 8 building types (which are a combination of the material type and the number of floors)
(Figure 4). The curves based on debris flow height were developed partly based on available
curves from the literature, and partly based on actual damage information from the 2003
event, combined with expert opinion. The maximum impact pressure found in the most
extreme event with the lowest return period (100-500y) was 35 KPa, which caused the total
destruction of several houses. Therefore, the maximum impact pressure for all other return
periods does not exceed 35 KPa and is considered a cut-off value.
Figure 4. Debris flow vulnerability curves for impact pressure (left) and debris height (right)
Local scale vulnerability assessment using photo-documentation
A logistic distribution was applied in order to model the debris flow vulnerability curve (see
Figure 5, Fella Data). The best fit parameters of the function (Equation 1) are given in Table
Where, P0 = initial population, K = limiting value of P0, r = growth rate, t = time. The
coefficient of determination of the Logistic distribution is R2 = 0.959.
International Conference
Analysis and Management of Changing Risks for Natural Hazards
18-19 November 2014 l Padua, Italy
Table 1 Logistic regression function
best fit parameters
Figure 5. Debris flow vulnerability curve for debris height
(Fella Data) in comparison with results from other similar
case studies
Population distribution, buildings inventory and value
In the study area, the inventory contains 4778 buildings. The building landuse types were
categorized in 16 classes. The residential and residential storage buildings are the most
frequent occupancy types, with 39.5% and 41.8% respectively (Figure 6). The percentage
distribution of the main occupancy types is reflected in the distribution of construction types:
generally, the residential structures are masonry constructions (or a combination of masonrywood or masonry-brick) (46.9%), whereas the majority of residential storage buildings (used
as sheds, cabins, or garages) are made out of wood (38.4%). The schools, hospitals,
municipal and governmental buildings are concrete or masonry constructions.
Figure 6. Percentage distribution of building occupancy types (left); variation in price for buildings
grouped per value range (right)
The building height regime reflects indirectly not only the use but also the design restrictions
given the seismic activity in the area. 60% of the buildings have only one level above the
ground surface, meanwhile only 0.29% of the total (14 buildings) has over 4 floors (up to 6
floors). These latter buildings are used mostly as multi-storey apartment buildings.
As each building in the inventory had a minimum and a maximum value assigned, it was
possible to calculate the variation in price for buildings grouped per value range. Figure 7
reflects the heterogeneity of buildings (in terms of use, material of construction, occupancy
type, etc.) for value categories of less than 10.000, 10.000 - 50.000, 500.000 – 800.000,
International Conference
Analysis and Management of Changing Risks for Natural Hazards
18-19 November 2014 l Padua, Italy
800.000 - 1.250.000 Euros, although for the latter two the number of buildings is significantly
lower compared with the former ones.
For the assessment of buildings exposure and risk we estimated the total worth of value
(minimum, maximum) potentially damaged in each commune. For example, for the
Malborghetto Valbruna commune, the total residential buildings value ranges between
approximately 190 – 275 million Euros followed by commercial ones with 130 – 190 million
The calculated number of people per residential building use for each commune and
scenario is presented in Table 2. The results show that in Malborghetto Valbruna, Pontebba
and Tarvisio the population increases whereas in Dogna, the calculated number of people for
both touristic and non-touristic season remains constant.
Table 2. Population distribution according to the building use taking into account two scenarios (A –
non-tourisic season, and B – touristic season)
Building use
Apartment building
1202 1202 1914 1914 83
Holiday apartment
Holiday home
1942 1942 104 104
Pharmacy & apartment building
Restaurant & apartment building
Shop & house
Shop & apartment building
1450 1490 1654 4503 5634 188 188
Total ISTAT (2014)
Vulnerability assessment represent an essential component of the for hazard risk analysis
framework. In this study, we have applied two methodological approaches in order to
quantify vulnerability of buildings to debris flow hazard using historical data and consequence
information of the August 2003 hydro-meteorological event in Fella River Basin.
In the first methodological approach, the probabilities of a regional scale debris flow runout
model were used to estimate intensity values expressed as impact pressures. The selection
of transfer functions as well as the distribution of the Flow-R spatial probabilities must be
further investigated in order to estimate the robustness of the model. Local scale dynamic
runout models can be used to compare the runout probabilities with debris flow heights in a
specific test area where more detailed data exists (see the work of Hussin et al., Abstract
The empirical vulnerability curve generated with the second methodological approach was
compared with other existing vulnerability curves such as the ones proposed by (Fuchs et al.,
2007, Akbas et al., 2009, Quan Luna et al., 2011) for similar buildings and processes. The
results for the Fella area show that the vulnerability values coincide in a good manner with
the literature curves; however, a direct comparison must be carefully interpreted as the
mathematical approaches and distributions are different. Another source of uncertainty is the
fact that intensity was assessed in different ways for each case study (for example, Quan
International Conference
Analysis and Management of Changing Risks for Natural Hazards
18-19 November 2014 l Padua, Italy
Luna et al. (2011) modelled the debris depths). Moreover, the inexactness of debris height
measurements represents a significant source of epistemic uncertainty.
In terms of buildings value and population distribution, it must be noted that building use and
spatial distribution of population during touristic and non-touristic season are bounded to high
variability. Nevertheless, comparing the official statistical data (INIS, 2014) with the
calculated population distribution per type of residential building (only for non-touristic
scenario), we can observe that the errors are negligible considering that the only input data
was the total number of residents per commune.
To improve the current work, more research must be invested in transferring results from
local to regional scale for both debris flow hazard modelling and vulnerability assessments.
As briefly discussed, the model has limitations and it requires further investigation. However,
the results can be readily applied in risk equations and provide to decision makers
information regarding the costs of events for different process intensities in the future or
under different development scenarios. As the implementation of risk reduction measures
(structural mitigation works) in the study area after the 2003 event can change the intensity
of future debris flows and thus assets‟ potential degree of loss and costs of reconstruction,
the current methodology can be further improved and used for the estimation of future
This work is a part of the CHANGES Project (Changing Hydro-meteorological Risks as
Analyzed by a New Generation of European Scientists), a Marie Curie Initial Training
Network, funded by the European Community‟s Seventh Framework Programme, FP7/2007–
2013, under Grant Agreement No. 263953. The authors kindly acknowledge the support of
the Civil Protection of Friuli-Venezia Giulia region and local authorities of Malborghetto
Valbruna municipality for sharing the photo-documentation and damage information related
with the August 2003 hydro-meteorological event.
AKBAS, S. O., BLAHUT, J. & STERLACCHINI, S. Critical assessment of existing physical vulnerability
approaches for debris flows. In: MALET, J. P., REMAITRE, A. & BOGAARD, T., eds.
International conference - Landslide processes: from geomorphologic mapping to landslide
modelling, 6–7 February 2009, 2009 Strasbourg, France. 229-233.
CNR-IRPI. 2014. AVI Project [Online]. Available: http://avi.gndci.cnr.it/en/archivi/frane_en.htm.
Quantification of model uncertainty in debris flow vulnerability assessment. Engineering
Geology, 181, 15-26.
FUCHS, S., HEISS, K. & HÜBL, J. 2007. Towards an empirical vulnerability function for use in debris
flow risk assessment. Natural Hazards and Earth System Science, 7, 495-506.
HORTON, P., JABOYEDOFF, M., RUDAZ, B. & ZIMMERMANN, M. 2013. Flow-R, a model for
susceptibility mapping of debris flows and other gravitational hazards at a regional scale.
Nat. Hazards Earth Syst. Sci., 13, 869-885.
ISPRA. 2014. The IFFI project (Inventory of Landslide Phenomena in Italy) [Online]. Available:
ITALIAN NATIONAL INSTITUTE OF STATISTICS (ISTAT) [Online]. Available: http://www.istat.it/
KAYNIA, A. M., PAPATHOMA-KÖHLE, M., NEUHÄUSER, B., RATZINGER, K., WENZEL, H. & MEDINACETINA, Z. 2008. Probabilistic assessment of vulnerability to landslide: Application to the
village of Lichtenstein, Baden-Württemberg, Germany. Engineering Geology, 101, 33-48.
OSSERVATORIO DEL MERCATO IMMOBILIARE (OMI), Agenzia Entrate (Italian Revenue Agency),
International Conference
Analysis and Management of Changing Risks for Natural Hazards
18-19 November 2014 l Padua, Italy
MALEK, Ž., SCOLOBIG, A. & SCHRÖTER, D. 2014. Understanding Land Cover Changes in the Italian
Alps and Romanian Carpathians Combining Remote Sensing and Stakeholder Interviews %M
doi:10.3390/land3010052 %U http://www.mdpi.com/2073-445X/3/1/52. Land %@ 2073445X, 3, 52-73.
PAPATHOMA-KÖHLE, M., KAPPES, M., KEILER, M. & GLADE, T. 2011. Physical vulnerability assessment
for alpine hazards: state of the art and future needs. Natural Hazards, 58, 645-680.
PAPATHOMA-KÖHLE, M., KEILER, M., TOTSCHNIG, R. & GLADE, T. 2012. Improvement of vulnerability
curves using data from extreme events: debris flow event in South Tyrol. Natural Hazards,
64, 2083-2105.
2011. The application of numerical debris flow modelling for the generation of physical
vulnerability curves. Nat. Hazards Earth Syst. Sci., 11, 2047-2060.
TOTSCHNIG, R. & FUCHS, S. 2013. Mountain torrents: Quantifying vulnerability and assessing
uncertainties. Engineering Geology, 155, 31-44.
TOTSCHNIG, R., SEDLACEK, W. & FUCHS, S. 2011. A quantitative vulnerability function for fluvial
sediment transport. Natural Hazards, 58, 681-703.
VARNES D. 1984. Landslide hazard zonation: a review of principles and practice. UNESCO, Paris