Download State-of-the-art in multi-risk assessment

Project number:
Project name:
Project acronym:
Theme:
Start date:
265138
New methodologies for multi-hazard and multi-risk
assessment methods for Europe
MATRIX
ENV.2010.6.1.3.4
Multi-risk evaluation and mitigation strategies
01.10.2010
End date: 30.09.2013 (36 months)
Deliverable
D5.1: State-of-the-art in multi-risk assessment
Version: Complete – under review
Responsible partner: AMRA
Month due:
M12
Month delivered:
M20
Primary authors(a):
ALEXANDER GARCIA-ARISTIZABAL and
WARNER MARZOCCHI
01.12.2011
_______________________________
_________________
Signature
Date
(
(a)
see the Acknowledgments section for a full list of people contributing to this document)
Reviewer:
GORDON WOO
01.12.2011
_______________________________
_________________
Signature
Date
Authorised:
Jochen Zschau
31.05.12
_______________________________
_________________
Signature
Date
1
Dissemination Level
PU
Public
Restricted to other programme participants (including the Commission
PP
Services)
Restricted to a group specified by the consortium (including the Commission
RE
Services)
CO
Confidential, only for members of the consortium (including the Commission
Services)
X
2
Abstract
The multi-risk concept refers to a complex variety of combinations of risk (i.e. various
combinations of hazards and various combinations of vulnerabilities) and for this reason it
requires a review of existing concepts of risk, hazard, exposure and vulnerability, within a
multi-risk perspective.
A multi-risk approach entails a multi-hazard and a multivulnerability perspective. The multi-hazard concept may refer to (1) the fact that different
sources of hazard might threaten the same exposed elements (with or without temporal
coincidence), or (2) one hazardous event can trigger other hazardous events (cascade
effects). On the other hand, the multi-vulnerability perspective may refer to (1) a variety of
exposed sensitive targets (e.g. population, infrastructure, cultural heritage, etc.) with
possible different vulnerability degree against the various hazards, or (2) time-dependent
vulnerabilities, in which the vulnerability of a specific class of exposed elements may
change with time as consequence of different factors (as, for example, wearing, the
occurrence of other hazardous events, etc.).
In this document we consider the main applications and research initiatives in the field of
multi-risk, considering both reports from European funded projects (and derived papers)
and also other international initiatives (reports and papers). From the bibliographic review
performed it emerges that most –if not all- of the initiatives on multi-risk assessment have
developed methodological approaches that consider the multi-risk problem in a partial
way, since their analyses basically concentrate on risk assessments for different hazards
threatening the same exposed elements. Within this framework, the main emphasis has
been towards the definition of procedures for the homogenization of spatial and temporal
resolution for the assessment of different hazards. For vulnerability instead, being a wider
concept, there is a stronger divergence over its definition and assessment methods;
considering physical vulnerability issues, a more or less generalized agreement on the use
of vulnerability functions (fragility curves) has been reached, which facilitate the application
of such a kind of multi-risk analysis, however, for other kinds of vulnerability assessment
(e.g. social, environmental, etc.) it is less clear how to integrate them within a multi-risk
framework.
In this framework the final multi-risk index is generally estimated as a simple aggregation
of the single indices estimated for different hazards; other approaches consider a single
hazard at a time and multiple exposed elements (e.g. buildings, people, etc.) for the
vulnerability, which are combined and weighted according to expert opinion and subjective
assignment of weights.
The choice of the methodology strongly depends on both the scale of the study and the
availability of information (for hazard and vulnerability assessment). Worthy of note, many
of the approaches found define theoretical frameworks for the multi-risk assessment that,
when applied to real cases, are generally simplified; this is due to the difficulty to obtain the
detailed information needed; it is also interesting to point out that many of the reports
discuss the importance of the interaction among hazards and cascading of events for a
fully multi-hazard perspective, however little effort has been made to define a rigorous
methodology.
Keywords: Multi-risk, literature review, state-of-the-art
3
Acknowledgments
The research leading to these results has received funding from the European
Community’s Seventh Framework Programme [FP7/2007-2013] under grant agreement n°
265138.
Different partners from the MATRIX project have collaborated in the preparation of this
report. For instance, the primary review of all the project reports, papers, etc. performed
here have been done by many people from different groups; here we include the entire list
of people who participated (alphabetical order of institutions):
- AMRA:
Alexander Garcia-Aristizabal, Warner Marzocchi;
- ASPINALL:
Gordon Woo;
- BRGM:
Arnaud Reveillere, John Douglas, Gonéri Le Cozannet;
- CEABN-ISA:
Francisco Rego, Conceicao Colaco;
- GFZ:
Kevin Fleming, Sergiy Vorogushyn;
- NGI:
Farrokh Nadim, Bjørn Vidar Vangelsten;
- TU-Delft:
Wouter ter Horst
4
Table of contents
Abstract ......................................................................................................................................... 3
Acknowledgments ........................................................................................................................ 4
Table of contents .......................................................................................................................... 5
List of Figures ............................................................................................................................... 7
List of Tables ................................................................................................................................. 9
1.
Introduction .......................................................................................................................... 10
2.
Summary of the main multi-risk initiatives (global, medium, local) ................................. 13
2.1
Large-scale approaches .............................................................................................................. 13
2.1.1
United Nations Development Program (UNDP, 2004) approach: The Disaster Risk
Index (DRI)............................................................................................................................................. 13
2.1.2
The World Bank approach: Natural Disaster Hotspots: A Global Risk analysis (e.g.
Dilley et al., 2005). ................................................................................................................................ 15
2.2
Medium-scale approaches .......................................................................................................... 18
2.2.1
EC-TIGRA project: “Multi-hazard risk assessment and zoning: an integrated
approach for incorporating natural disaster reduction into sustainable development” (e.g. Del
Monaco et al., 1999)............................................................................................................................. 18
2.2.2
The European Multi-Hazard Risk Assessment Project (TEMRAP), (European
Commission, 2000) ............................................................................................................................... 19
2.2.3
The Spatial Effects and Management of Natural and Technological Hazards in
Europe (ESPON) project (Schmith-Thomé, P. Ed., 2005) ............................................................. 22
2.2.4
The Applied multi-risk mapping of natural hazards for impact assessment
(ARMONIA) project (ARMONIA, 2007). ............................................................................................ 26
2.2.5
The Federal Emergency Management Agency initiatives: “Multi-hazard identification
and risk assessment” (FEMA, 1997) and “Understanding your risks: identifying hazards and
estimating losses” (FEMA, 2001). ...................................................................................................... 34
2.2.6
2.3
The natural- and conflict-related hazards in Asia-Pacific project (OCHA, 2009) ........ 38
Regional’ to ‘Local’ scale approaches ....................................................................................... 40
2.3.1
The “Natural Risk Assessment” (NARAS) project (e.g. Marzocchi et al., 2009) ......... 40
2.3.2
The “Risque Naturel Transverse” (RISK-NAT) project (e.g., Carnec et al., 2005;
Douglas 2005, Douglas 2007) ............................................................................................................ 45
2.3.3
The comparative multi-risk assessments for the city of Cologne, Germany (e.g.
Grunthal et al., 2006; Kleist et al., 2006; Merz and Thieken, 2009) .............................................. 45
2.3.4
The regional-level multi-risk project in the Piedmont region, Italy: “A methodological
approach for the definition of multi-risk maps at regional level: first application” (Carpignano et
al., 2009) ................................................................................................................................................ 49
2.3.5
The regional-level initiative of integration of natural and technological risks in
Lombardy region, Italy (e.g. Lari et al., 2009) ................................................................................... 53
5
2.3.6
A conceptual model for “Reconsidering the risk assessment concept: Standardizing
the impact description as a building block for vulnerability assessment” (Hollenstein, 2005) .. 54
2.3.7
The Cities project for geohazards in Australian urban communities (e.g. Grander et
al., 1999; Granger and Haine, 2001); ................................................................................................ 55
2.3.8
A multi-hazard risk assessment for the Turrialba city, Costa Rica (e.g., Van Wasten
et al., 2002); ........................................................................................................................................... 59
2.3.9
The Central American Probabilistic Risk Assessment (CAPRA) approach
(http://www.ecapra.org);....................................................................................................................... 62
2.3.10 The ‘Regional RiskScape’ project in New Zealand: “Quantitative multi-risk analysis for
Natural hazards: a framework for multi-risk modelling.” (Schmidt et al., 2011) .......................... 67
2.3.11
3.
4.
Multi-risk initiatives in volcanic areas ................................................................................. 70
Summary of the main results: general maps, risk indices and risk curves ..................... 73
3.1
General considerations ................................................................................................................ 73
3.2
Scale constraints ........................................................................................................................... 74
3.3
Presentation of results: general risk maps, risk curves and risk indices. ............................. 74
3.4
Final remarks, limitations and gaps ........................................................................................... 77
References ........................................................................................................................... 79
6
List of Figures
Figure 1.1 Yearly number of (a) multi-hazard and (b) multi-risk related papers published in indexed
journals ........................................................................................................................................................... 10
Figure 2.1 physical exposure and relative vulnerability to floods in the period 1980-2000 ............... 15
Figure 2.2 Global distribution of flood risk: mortality (a), total economic loss (b), and economic loss
as proportion of GDP density (c) (source: Dilley et al., 2005)................................................................ 18
Figure 2.3 Possible approaches to develop a global economic risk ..................................................... 20
Figure 2.4 (a) aggregated hazard map (based on 15 hazard indicators); (b) Integrated vulnerability
map; (c) aggregated risk .............................................................................................................................. 24
Figure 2.5 Integrated vulnerability index ................................................................................................... 25
Figure 2.6 ARMONIA function between hazard types intensity (x-axis) vs. average damage of
exposed elements (y-axis) according to scale of analysis ..................................................................... 28
Figure 2.7 ARMONIA function between hazard types intensity (x-axis) vs. probability of damage for
different categories of exposed elements (y-axis) according to scale of analysis .............................. 28
Figure 2.8 (a) Hypothetical vulnerability functions relating the intensity of a hazard to a risk metric;
(b) Aggregating risk from multi-hazards to assess the average annual risk (source: ARMONIA,
2007). Note: the above curves are hypothetical probability versus risk metric curves to illustrate the
method by which risk metrics from different natural hazards can be combined ................................. 30
Figure 2.9 ARMONIA methodology for producing a risk index for a particular hazard ...................... 31
Figure 2.10 Diagram to show hazard levels and indices as a function of probability and intensity . 31
Figure 2.11 Flow chart showing the method used for producing vulnerability and consequence
indices............................................................................................................................................................. 32
Figure 2.12 Methodology employed in the ARMONIA Decision Support System (DSS) .................. 33
Figure 2.13 HAZUZ-MH multi-hazard analysis levels ............................................................................. 36
Figure 2.14 Final output of the multi-risk analysis done for the city of Cologne, Germany (Grünthal
et al., 2006): Risk curves of the hazards due to windstorms, floods and earthquakes for the city of
Cologne for losses concerning buildings and contents in the sectors private housing, commerce
and industry (reference year: 2000). .......................................................................................................... 46
Figure 2.15 Sub-tasks for each step of the risk assessment process .................................................. 47
Figure 2.16 (from Merz & Thieken, 2009). The left figure is similar to one from Grünthal et al. (
2006) for floods with, in addition, a cluster of curves corresponding to the evaluation and
propagation of uncertainties through the loss estimation. The contribution of different steps is
illustrated on the right figure. ....................................................................................................................... 49
Figure 2.17 The vulnerability model proposed by Carpignano et al. (2009)........................................ 50
Figure 2.18 How the combination of scenarios (i) and damage indicators (j) determine the risk
indices Rj,i as used by Carpignano et al. (2009). .................................................................................... 51
Figure 2.19 Example of an aggregated risk map edited by the Civil Protection Department for
delivery infrastructures which are exposed to five hazard scenarios: seismic event, industrial
accident, flood, landslide and forest fire (Carpignano et al. 2009). ....................................................... 53
Figure 2.20 Comparison of percent building and contents damage from earthquake and cyclone
wind ................................................................................................................................................................. 58
Figure 2.21 Comparison of percent building and contents damage from storm tide inundation and
flood ................................................................................................................................................................ 59
Figure 2.22 Specific risk curve for flooding (left) and seismic (right). X-axis is annual exceedance
probability; Y-axis is estimated damage in Costa Rican currency ........................................................ 62
Figure 2.23 Modelling method of hazards based on triggering events ................................................ 64
7
Figure 2.24 Theoretical framework and model for holistic approach of disaster risk (from Carreño et
al., 2007) ........................................................................................................................................................ 67
Figure 2.25 Example output for a risk assessment for the Hawke’s Bay area using the RiskScape
software. (a) 1931 historical earthquake event applied to calculate building losses in terms of
reconstruction costs (scenaro); (b) 1,000-year design storm event applied to calculate building
damages, here expressed as damage ratio (% destroyed); (c) Flood inundation from a Tutaekuri
Meeanee river breach scenario applied to calculate building damages, here expressed as damage
ratio (% destroyed). ...................................................................................................................................... 69
Figure 2.26 Frequency distributions (% of affected area exceeding certain loss) of total building
damage (as reconstruction costs per km2) for the earthquake, flood and storm scenarios for
Hawke’s Bay building assets. The earthquake scenario has more higher damages (~70% of
damages>$10,000/km2; ~5% of damages>$1,000,000/km2); the flood scenario has more lower
damages (for ~50% of the area, damages are negligible; for the rest of the area, the damages vary
exponentially between $100,000/km2 and $100 Mio/km2); the storm scenario has similar
frequency characteristics compared to the earthquake scenario with higher frequency of damages
between $10,000/km2 and $1Mio/km2 (~80% of damages>$10,000/km2, almost 100% of
damages < $10 Mio/km2) ............................................................................................................................ 70
Figure 2.27 Diagram of the vulnerability assessment principle used in Thierry et al., 2006 ............. 72
Figure 3.1 Summary of the most important typologies of results of multi-risk methodologies found
in literature: (a) example of global distribution of flood risk (expected mortality, from Dilley et al.,
2005) ), for details see Figure 2.2 in this document; (b) example of an aggregated risk map from
the ESPON methodology (from Schmith-Thomé, P. Ed., 2005); for details see Figure 2.4 in this
document; (c) combination of scenarios (i) and damage indicators (j) to determine the risk indices
Rj,i (from Carpignano et al., 2009), for details see Figure 2.18 and Figure 2.19; (d) diagram of a
proposed ARMONIA function between hazard types intensity (x-axis) vs. average damage of
exposed elements (y-axis) according to scale of analysis (from ARMONIA, 2007, for details see
Figure 2.6 of this document); (e) comparison of risk curves for the hazards due to windstorms,
floods and earthquakes (from Grünthal et al., 2006), for details see Figure 2.14 in this document;
(f) risk ranking (from Marzocchi et al., 2009 [Note: the sketch of the risk curves is from ARMONIA,
2007]).............................................................................................................................................................. 77
8
List of Tables
Table 2.1 Synthesis of the hazard assessment for floods ...................................................................... 19
Table 2.2 Typological classification of natural hazards in Europe following the TEMRAP approach
......................................................................................................................................................................... 21
Table 2.3 Potential spatial field of application (X mark) of suggested procedures following the state
of the art on multi-hazard risk assessment ............................................................................................... 29
Table 2.4 Risk index with the same weighting factor for all hazards .................................................... 41
Table 2.5 Procedures used for single-risk assessment (hazard and vulnerability approaches) in the
Casalnuovo case-study of NARAS project ............................................................................................... 44
Table 2.6 parameters used for the hazard assessment (Grunthal et al., 2006) ................................. 48
Table 2.7 Summary of vulnerability assessment methodologies .......................................................... 48
Table 2.8 list of generic components that may characterize the hazard impact intensity (I) ............ 54
Table 2.9 Considered hazards in the Cities project for geohazards in Australian urban communities (e.g.
Grander et al., 1999) ....................................................................................................................................... 56
Table 2.10 Intensity parameters considered of the different hazards .................................................. 64
9
1. Introduction
The main objective of Task 5.1 of the WP5 of the MATRIX project is to perform a review
and to summarize the existing methods and procedures for multi-risk assessment. The
goal is to establish the state-of-the-art and to identify both the gaps in knowledge and the
possible shortcomings of existing methodologies. This initiative is strongly linked to the
Task 3.1 of the WP3 (review of multi-hazard assessment procedures) that is mostly
focused on reviewing the state-of-the-art of multi hazard assessment, with particular
attention to the work undertaken in previous European projects. From those two tasks,
two deliverables are expected as products: D5.1 (State-of-the-art in multi-risk assessment)
and D3.1 (Review of existing procedures: review of the existing procedures for multi-risk
assessment (with emphasis on multi-hazard procedures)). Given that a multi-risk
approach entails a multi-hazard perspective, we have found that the state-of-the-art on
both multi-hazard and multi-risk procedures in general share common sources. For this
reason, it was decided to keep within deliverable D5.1 the most general discussion on
existing multi-risk procedures (and then the correlated multi-hazard approaches), as within
D3.1 we concentrate on the specific strategies adopted as ‘multi-hazard’ analyses.
The first activity of Task T5.1 (in coordination with T3.1) was to collect the information
about the most important past projects and scientific publications in which the multi-risk
(and usually the implicit multi-hazard) assessment problem has been approached. For
instance, the growing awareness of the need to develop multi-risk approaches has led to
the development of different projects in Europe and in different countries with the aim of
realizing tools and procedures for successful land planning and management of territory,
trying to homogenize existing methodologies within a unique approach. This growing
interest on multi-risk procedures is also evident in the scientific literature, which is
evidenced if we count the number of publications in indexed journals in which the
arguments “multi-hazard” and “multi-risk” have been considered (See Figure 1.1).
(a)
(b)
2010
2009
year of publication
Year of publication
2010
2008
2007
2006
2005
2003
2008
2006
2004
2001
1999
1997
1997
1984
1982
0
2
4
6
8
Number of published papers (ISI journals)
Multi-hazard
10
0
2
4
6
8
10
12
Number of published papers (ISI journals)
Multi-risk
Figure 1.1 Yearly number of (a) multi-hazard and (b) multi-risk related papers published in indexed journals
From a conceptual point of view, the multi-risk concept refers to a complex variety of
combinations of risk (i.e. various combinations of hazards and various combinations of
vulnerabilities) and for this reason it requires a review of existing concepts of risk, hazard,
exposure and vulnerability, within a multi-risk perspective (e.g. Carpignano et al., 2009). A
multi-risk approach entails a multi-hazard and a multi-vulnerability perspective. The multihazard concept may refer to (1) the fact that different sources of hazard might threaten the
same exposed elements (with or without temporal coincidence), or (2) one hazardous
event can trigger other hazardous events (cascade effects). On the other hand, the multivulnerability perspective may refer to (1) a variety of exposed sensitive targets (e.g.
population, infrastructure, cultural heritage, etc.) with possible different vulnerability degree
10
against the various hazards, or (2) time-dependent vulnerabilities, in which the vulnerability
of a specific class of exposed elements may change with time as consequence of different
factors (as, for example, wearing, the occurrence of other hazardous events, etc.).
In this document we consider the main applications and research initiatives in the field of
multi-risk, considering both reports from European funded projects (and derived papers)
and also other international initiatives (reports and papers). For instance, a series of
European funded projects, as well as different practical applications in different states
worldwide, have been found. The different approaches may be distinguished, arbitrarily,
as a function of the considered scale (note that some approaches may be traverse
different scales):
-
‘Large’ scale approaches: at this level, multi-risk procedures generally emphasize
the areas with the relative highest risks, highlighting wide (country- or continentsize) areas that are more affected by different hazards that others. These largescale maps are generally based on a simple hazard aggregation (a multi-hazard
perspective) and generally represent a simple risk indicator. Within this group we
can find the multi-hazard/risk (worldwide) maps developed by the following:
o The United Nations Development Program (UNDP, 2004), in which the risk
indicator is represented as an average risk of deaths;
o The Munich Re maps (http://www.munichre.com), in which hazards are
described worldwide for different intensities of the phenomena, and the risk
indicator is expressed as average or potential losses;
o The World Bank approach, (Dilley et al., 2005), in which a series of risk
indices are calculated and presented as hotspots in maps, as for example,
the multi-hazard mortality risk hotspots.
-
‘Medium’ scale approaches: at this level we can include many of the European
initiatives on multi-risk assessment. The most important past projects we can find
are as follows:
o the EC-TIGRA project (e.g. Del Monaco et al., 1999): “Multi-hazard risk
assessment and zoning: an integrated approach for incorporating natural
disaster reduction into sustainable development”;
o The TEMRAP project (European Commission, 2000): “The European MultiHazard Risk Assessment Project”; the ESPON project (Schmith-Thomé, P.
Ed., 2005): “The Spatial Effects and Management of Natural and
Technological Hazards in Europe”;
o The ARMONIA project (2007): “Applied multi-risk mapping of natural hazards
for impact assessment”.
Different international projects can be also included within this group, as for
example: the Federal Emergency Management Agency initiative (e.g. FEMA,
1997): “Multi-hazard identification and risk assessment”, a regional- to localscale approach in which an interesting decision support system (HAZUS) was
developed; and the natural- and conflict-related hazards in Asia-Pacific project
(OCHA, 2009), in which a risk index was developed considering for its
calculation seven (natural and man-made) hazards.
11
-
‘Regional’ to ‘Local’ scale approaches: at this level, different specific initiatives
can be found in literature. In Europe, for example, we found:
o The European-funded NARAS project (2006): “Natural Risk Assessment”
(e.g. Marzocchi et al., 2009), in which a general discussion of multi-risk
assessment with an application to the Casalnuovo municipality, Italy, was
performed;
o The RISK-NAT (Risque Naturel Transverse) project (e.g., Carnec et al.,
2005; Douglas 2005, Douglas 2007);
o The comparative multi-risk assessments for the city of Cologne, Germany
(e.g. Grunthal et al., 2006; Kleist et al., 2006; Merz and Thieken, 2009) within
the DFNK (German Research Network Natural Disasters) project;
o A regional-level multi-risk project in the Piedmont region, Italy (Carpignano et
al., 2009);
o The regional-level initiative of integration of natural and technological risks in
Lombardy region, Italy (e.g. Lari et al., 2009);
o The distributed framework for multi-risk assessment, an initiative in the
framework of the MEDIGRID project, with the aim of to model the effects of
forest fires on hydrology and sediment yield (e.g. Bovolo et al., 2009);
Among other international (extra European) initiatives we found: the Cities
project for geohazards in Australian urban communities (e.g. Grander et al.,
1999); a multi-hazard risk assessment for the Turrialba city, Costa Rica (e.g.,
Van Wasten et al., 2002); the Central American Probabilistic Risk Assessment
(CAPRA) approach (http://www.ecapra.org); and the ‘Regional RiskScape’
project in New Zealand (e.g. Schmidt et al., 2011). Finally, three initiatives of
kind of multi-risk assessment but considering just volcanic areas (a single or
multiple volcanic sources generating different kinds of volcanic-related hazards)
have been found: the “Explosive Eruption Risk and Decision Support for EU
Populations Threatened by Volcanoes” (EXPLORIS) project (e.g., Hincks et al.,
2006); a volcanic risk ranking initiative for the Auckland region, New Zealand
(e.g. Magill and Blong, 2005a,b) and the GRINP project at Mount Cameroon
volcano (e.g. Thierry et al., 2008);
12
2. Summary of the main multi-risk initiatives (global, medium, local)
In this section, we perform a summary of some of the main multi-risk methodologies
described in the introductory part. Following the same division scheme, they are
presented following the large, medium and regional to local scale classification.
2.1
Large-scale approaches
2.1.1 United Nations Development Program (UNDP, 2004) approach: The
Disaster Risk Index (DRI).
The pilot DRI, presented in the UNDP (2004) report, enables the measurement and
comparison of relative levels of physical exposure to hazard, vulnerability and risk between
countries. The DRI is the concept that disaster risk is not caused by hazardous events per
se, but rather is historically constructed through human activities and processes. As such
the risk of death in a disaster is only partially dependent on the presence of physical
phenomenon such as earthquakes, tropical cyclones and floods. In the DRI, risk refers
exclusively to the risk of loss of life and excludes other facets of risk, such as risk to
livelihood and to the economy.
The formula used for modelling risk combines its three components; risk is a function of
hazard occurrence probability, the element at risk (population) and vulnerability. The
equation below was made for modelling disaster risk:
R = H • Pop • Vul (approach 1)
Where:
R is the risk (in terms of number of killed people);
H is the hazard, which depends on the frequency and strength of a given hazard;
Pop is the population living in a given exposed area;
Vul is the vulnerability and depends on the socio-political-economical context of the
population.
Hazard multiplied by the population was used to calculate physical exposure.
R = PhExp • Vul (approach 2)
Where PhExp is the physical exposure, i.e. the frequency and severity multiplied by
exposed population. Physical exposure was obtained by modelling the area affected by
each recorded event. Event frequency was computed by counting the number of events for
the given area, divided by the number of years of observation (in order to achieve an
average frequency per year). Using the area affected, the number of people in the
exposed population was extracted using a Geographical Information System (GIS). The
population affected multiplied by the frequency of a hazard event for a specified magnitude
provided the measure for physical exposure.
In the DRI, vulnerability refers to the different variables that make people less able to
absorb the impact and recover from a hazard event. These may be economic (such as
lack of reserves or low asset levels); social (such as the absence of social support
mechanisms or weak social organization); technical (such as poorly constructed, unsafe
13
housing); and environmental (such as the fragility of ecosystems). Due to the hazard
specificity of people’s vulnerability, it is not conceptually possible to arrive at a global multihazard indicator of vulnerability. Rather the vulnerability indicators suggested by the DRI
are always hazard specific.
We can calculate the relative vulnerability of a country to a given hazard by dividing the
number of people killed by the number exposed. When more people are killed with respect
to the number exposed, the relative vulnerability to the hazard in question is higher.
To estimate the total risk:
RiskTot =Σ(RiskFlood + RiskEarthquake + RiskVolcano + RiskCyclone + ...Riskn )
The multi-hazard risk for a country required calculating an estimate of the probability of the
occurrence and severity of each hazard, the number of persons affected by it, and the
identification of the population’s vulnerability and coping capacities. This is very ambitious
and not achievable with present data constraints. However the aim is to provide an
approach built on existing data that will be refined in subsequent runs of the DRI. (See
study cases)
-
-
-
-
Scale: The DRI has been developed with a global level of observation and a
national level of resolution, allowing comparison between countries.
Considered hazards: earthquakes, tropical cyclones and floods. They also tried to
work with droughts however due to the lack of reliable data, this hazard was
excluded from the final results. However, they have included it in the final multi-risk
equation.
Added value: The DRI enables the calculation of the average risk of death per
country in large- and medium-scale disasters associated with earthquakes, tropical
cyclones and floods, based on data from 1980 to 2000. It also enables the
identification of a number of socio-economic and environmental variables that are
correlated with risk to death and which may point to causal processes of disaster
risk. In the DRI, countries are indexed for each hazard type according to their
degree of physical exposure, their degree of relative vulnerability and their degree
of risk.
Limitations: It represents a synoptic methodology principally addressed to global
policies with very low reliability at the local scale. It considers a limited number of
hazards and it does not consider possible cascade effects. The DRI only represents
the primary hazard events as recorded in global disaster databases, even when in
some cases the majority of loss may be associated with a range of different hazard
types triggered by the primary event.
Uncertainties: not discussed
An example of the results of this approach is shown in Figure 2.1, which shows the
physical exposure and relative vulnerability to floods in the period 1980-2000 (source:
UNDP, 2004)
14
Figure 2.1 physical exposure and relative vulnerability to floods in the period 1980-2000
2.1.2 The World Bank approach: Natural Disaster Hotspots: A Global Risk
analysis (e.g. Dilley et al., 2005).
The Natural Disaster Hotspots (NDH) project generated a global disaster risk assessment
and a set of more localized or hazard-specific case studies. In the report the global risks of
two disaster-related outcomes is assessed: mortality and economic losses. They estimate
risk levels by combining hazard exposure with historical vulnerability for two indicators of
elements at risk: (1) gridded population and (2) Gross Domestic Product (GDP) per unit
area, for six major natural hazards: earthquakes, volcanic eruptions, landslides, floods,
drought, and cyclones. By calculating relative risks for each grid cell rather than for
countries as a whole, it is possible to estimate risk levels at sub-national scales. A set of
accompanying case studies, available separately, explores risks from particular hazards or
for localized areas in more detail, using the same theoretical framework as the global
analysis.
The global analysis undertaken in this project is limited by issues of scale as well as by the
availability and quality of data (e.g. for a number of hazards they had only 15- to 25-year
records of events for the entire globe and relatively crude spatial information for geolocating these events; on the other hand, data on historical disaster losses, and particularly
on economic losses, are also limited). However, it is discussed that while the data are
inadequate for understanding the absolute levels of risk posed by any specific hazard or
combination of hazards, they are adequate for identifying areas that are at relatively higher
single- or multiple-hazard risk; for example, they can identify those areas that are at higher
risk of flood losses than others and at higher risk of earthquake damage than others or at
higher risk of both. Within the constraints summarized above (among others), in this
project three indices of disaster risk were developed:
-
Disaster-related mortality risk, assessed for global gridded population;
Risk of total economic losses, assessed for global gridded GDP per unit area; and
15
-
Risk of economic losses expressed as a proportion of the GDP per unit area and for
each grid cell.
Since the objective of this analysis is to identify hotspots where natural hazard impacts
may be large, it includes the large proportion of the Earth’s land surface that is sparsely
populated and not intensively used. They have therefore chosen to mask out grid cells with
population densities less than 5 persons per square kilometre (cells with less than about
105 residents) and without significant agriculture. Even if all residents of such cells were
exposed and highly vulnerable to a hazard, total casualties would still be relatively small in
absolute terms, and the potential agricultural impact would be limited. Masking these cells
reduces data processing requirements and ensures that the large number of very low risk
cells do not dominate the results. In addition, hazard reporting and frequency data are
likely to be poorest in rural, sparsely populated areas, so masking could help to reduce
anomalies caused by poor data. Next, they used historical losses as recorded in EM-DAT
(the ‘International Disaster Database’) across all events from 1981-2000 for each hazard
type to obtain mortality and economic loss weights for each hazard for each region for four
economic wealth classes within regions. The weights are an aggregate index of relative
losses within each region and country wealth class for each hazard over the 20-year
period.
The risk assessment procedure for both mortality and economic losses can be
summarized as follows:
-
-
-
The appropriate measure of total global losses from 1981-2000 from EM-DAT (in
the mortality case, the number of fatalities) is extracted by hazard h: Mh.
Using the GIS data on the extent of each hazard, they compute the total population
estimated to live in the area affected by that hazard in the year 2000: Ph.
A simple mortality rate for the hazard is computed: rh = Mh/Ph. Assuming that the
1981-2000 period was representative, this rate is an estimate of the proportion of
persons killed during a 20-year period in the area exposed to that hazard. Since the
numbers are very small, they are expressed per 100,000 persons in 2000. Future
revisions of the index could construct a mortality rate for the 20-year period based
on annual rates which are computed using yearly mortality and population figures.
As the results are intended only as an index of disaster risk, however, they believe
that the computational simplification of using only end-of-period population is
justified.
For each GIS grid cell i that falls into a hazard zone h, the location-specific
expected mortality is computed by multiplying the global hazard-specific mortality
rate by the population in that grid cell: Mhi = rh*Pi. It is done for all six hazards, then
a mortality weighted multi-hazard index value is computed for each grid cell:
. This first estimate represents an un-weighted index value that
assumes that mortality rates are globally uniform and that hazard severity has no
influence on the relative distribution of mortality.
Denoting the various combinations of region and country-wealth status by j, then
the estimated mortality in a given grid cell is Mhij=rhj*Pi.
The global hazard data compiled for the analysis measures the degree of hazard in
terms of frequency in most, although not all, cases. The various degree of hazard
measures are used to redistribute the total regional mortality from EM-DAT across
16
-
-
-
the grid cells in the area of the region exposed to each hazard. For example, if a
grid cell were hit four times by a severe earthquake during the 20-year period, the
regional mortality rate is multiplied by four to yield an accumulated mortality for that
grid cell. More generally, if the degree of hazard measure is denoted by W, and
assuming that the weighting scheme is identical across region/wealth-class
combinations j, the accumulated mortality in the grid cell is: M’hij = rhj*Whi*Pi.
Since the degree of hazard is not always measured on the same scale across
hazards, simply adding up the resulting values would result in an index that could
be unduly dominated by a hazard that happens to be measured on a scale with
larger values. In the methodology therefore they deflate the weighted hazardspecific mortality figures uniformly, so that the total mortality in each region adds up
to the total recorded in EM-DAT. The resulting weighted mortality from hazard h in
grid cell i and region/wealth-class combination j is thus:
,
where n is the number of grid cells in the area exposed to hazard h.
A mortality-weighted multi-hazard disaster risk hotspot index can be then calculated
as the sum of the adjusted single-hazard mortalities in the grid cell across the six
hazard types:
.
To avoid literal interpretation of the multi-hazard disaster risk hotspot index as the
number of persons expected to be killed in a 20-year period and in recognition of
the many limitations of the underlying data, they convert the resulting measure into
an index from one to ten using a classification of the global distribution of unmasked
grid cell values into deciles. Figure 2.2 is an example of the results for the global
distribution of flood risk.
17
Figure 2.2 Global distribution of flood risk: mortality (a), total economic loss (b), and economic loss as
proportion of GDP density (c) (source: Dilley et al., 2005)
2.2
Medium-scale approaches
2.2.1 EC-TIGRA project: “Multi-hazard risk assessment and zoning: an
integrated approach for incorporating natural disaster reduction into
sustainable development” (e.g. Del Monaco et al., 1999).
The TIGRA project aimed to develop an integrated approach considering the state of
natural environment in terms of climate (floods, landslides, coastal, etc.) and geophysical
(earthquakes, volcanic eruptions, etc.) events, the state of anthropogenic system
(exposure and related vulnerability), the mutual relationships, and the mitigation strategies.
The aim of the project was to homogenize existing methodologies on individual perils
within a unique approach. It was developed as a feasibility project for understanding the
possibility to realize tools and procedures for a successful land planning and management
of territory. For any typology of hazard, a specific summary of the methodology has been
provided, demonstrating how different natural phenomena may be constrained in a
process grid. The assessment of each considered hazard is synthesized using common
steps, identifying the event typology, the input information (scale, data inventory,
predisposing areas, triggering events, long term evaluation, monitoring and geoindicators), the modelling procedures, and the output (the kind of assessment or
‘prediction’); Table 2.1 is an example of the proposed synthesis for the case of floods.
18
Table 2.1 Synthesis of the hazard assessment for floods
As an important added value, it is possible to define multi-hazard risk assessment by
means of economic indices reporting the expected economic losses resulting from each
individual procedure applied to single hazards. At a regional scale (e.g. municipality area
as elementary cell) the proposed result can be defined in a “susceptibility risk
assessment”, i.e. the possibility to suffer economic losses from natural events (0 – 1); at a
local scale (elementary cell represented by single exposed objects such as single or
aggregated buildings or infrastructures) risk is represented by expected economic losses
for each exposed element.
2.2.2 The European Multi-Hazard Risk Assessment Project (TEMRAP),
(European Commission, 2000)
The aim of The European Multi-hazard Risk Assessment Project, TEMRAP, was to
develop an integrated methodology on multi-hazard and risk assessment, on the basis of
different experiences carried out in several European countries on natural disasters. The
project was mainly focused on the identification of natural hazards, profile hazards,
evaluation of their potential consequences, and mapping/zoning data on a GIS based
system. The approach was to be undertaken at a regional scale in Western Italy
(macrozoning) and refined in the urban area of Genoa (microzoning). This approach is
based on a comprehensive investigation of the environment and human structures and
infrastructures in order to define the most suitable mitigation strategy for any specific
target. From the analysis of potential users it clearly arose that data input, methodologies
and output are strongly dependent on final users demand; in such a way three main areas
were defined:
-
Federal/regional land planning and protection: susceptibility approach;
Local planning and protection: hazard approach;
Site mitigation: site engineering approach.
The TEMRAP project was focused on underlining a common strategy among different
natural events, although the methodology for internal data elaboration and manipulation is
clearly different. The most reliable models can be implemented for climate-related hazards
(i.e. floods, landslides, strong winds) whereas geophysical hazards need to be
investigated individually.
19
TEMRAP Multi-hazard/risk assessment approach: The investigation developed in the
project highlighted the difficulties in defining a simple approach given by overlapping
different hazard probability maps. Even if an inventory map synthesizing all phenomena
can be easily produced, at both local and regional scales, many conceptual problems arise
when different probabilities of occurrence in the future (hazard) are assessed for the same
specific elementary unit of territory. Nevertheless, it is suggested that the only way of
producing a synoptic view of expected losses can be expressed in terms of economic
indices that summarize the contribution of individual risk procedures. Figure 2.3 provides
one of the possible approaches to develop a global economic risk.
Figure 2.3 Possible approaches to develop a global economic risk
In practical terms, the expected potential unitary losses are expressed in the following
way:
where: is the typology of phenomenon; the probability of occurrence; the composed
vulnerability;
a coefficient of exposure to hazard, and
the expected potential unitary
losses.
At a regional scale the above formula can be defined in terms of per capita product (pcp),
where the expected losses are computed as:
Where
are the per capita expected losses, and, consequently, multiplying by
population we can obtain the total loss by municipality.
The first step of the process is related to the definition of hazard, namely the event
probability at a local scale (or susceptibility at a regional scale); this is performed using
standard methodologies for each typology of hazard (separately), even following the same
chain to get the same type of result (probability of occurrence or susceptibility of
occurrence). In general, all the natural hazards were grouped in two main categories such
as geophysical dynamics and climate dynamics, within which they can be further
characterized by rapid onset and slow onset parameters (e.g. see Table 2.2).
20
Geophysical
and
Meteo-climate
Rapid onset
Slow onset
Volcanic eruption
Heavy rainfall
Strong wind
Landslide
Flood
Earthquake
Karst
Fire
Coastal dynamic
Subsidence
Sea level change
Table 2.2 Typological classification of natural hazards in Europe following the TEMRAP approach
In TEMRAP, it is stated that environmental risk studies require a typical multidisciplinary
approach, as a fundamental tool of analysis to recognize the link among various hazard
processes and types (e.g. common triggering factors). Only through multilevel approaches
is it possible to act over the original causes in an organic and effective way, in the attempt
to resolve the product of the single effects separately from the others. In the TEMRAP
project the analysis of the distinct triggering factors affecting all potential hazards, in a
study area, has permitted users to identify them as the key to define the different levels of
risk; such an approach allows one to overcome the difficulty to define multi-hazard, as an
early step leading to risk analysis. Seismic events, for example, are often followed by
landslides that can produce greater damage than that caused by earthquakes. Therefore,
a correct land planning strategy should consider a seismic microzoning coupled with a
study of earthquake-induced landslides. Another example is given by heavy rainfall: such
events generally affect mountainous areas, from river catchments areas to alluvial fans
and coastal zones. The multi-risk approach applied in the study area has stressed some
common triggering factors for different natural disaster events.
For the vulnerability, the main propose in the context of TEMRAP project was the selection
of crucial indicators to help assess social and economic vulnerability. Social and economic
parameters referred in very general terms to the response of a human system when hit by
extreme natural events. The impact of this response can be considered as the ability (or
inability) to cope with
emergency management, and in terms of resources for
reconstruction, for re-establishing normal patterns of life. In the first two phases, the goal
was to assess what the reaction of the social and economic system would be, how well or
how poorly would be the confrontation with physical damage and infrastructure disruption.
In the last phase, parameters might make explicit existing opportunities for rebuilding. The
TEMRAP methodological approach for social an economic vulnerability may be
summarized as:
-
-
To identify a limited set of social and economic parameters to assess how sound is
the social and economic structure of a region at risk for emergency response (e.g.
relative number of the very old and very young -the most difficult to evacuate in
case of need-; fluctuation of the population during the year (e.g. coinciding with
holidays); the ratio between concentrated, large settlements and dispersed ones
(this is a very important variable with respect to emergency operations).
To identify another set of social and economics parameters to assess existing
human resource and economic opportunities for reconstruction and for reducing
21
-
pre-event vulnerabilities (e.g. demographic trend over the last 30 years -so as to
assess if the area is going to be abandoned or if it is developing-; ratio of the active
population with respect to the total; number of factories of different dimensions with
respect to the total; number of the provinces and the regions -so as to have an idea
of the robustness of economic activities in the different areas-; ratio of owned
dwellings; distribution of workers in economic sectors, etc.).
To identify set of parameters refers to the use and occupancy of buildings (e.g. ratio
between occupied and not occupied houses; ratio of non-occupied houses used as
vacation houses; age and quality of buildings).
As final remarks, in the TEMRAP methodology multi-risk analysis is theoretically possible
by integrating the impact of the various types of hazard within the socio-economic setting
of the study area; multi-hazard mapping is heavily constrained by typical dimensions of the
natural event, data availability, and methods of analysis and scales of representation,
however, multi-hazard analysis is possible only for disasters triggered by common factors.
2.2.3 The Spatial Effects and Management of Natural and Technological Hazards
in Europe (ESPON) project (Schmith-Thomé, P. Ed., 2005)
In the context of the European Spatial Planning Observation Network (ESPON) not all
hazards are relevant. Therefore relevant hazards were selected according to specified risk
criteria. The selection of hazards was done in three steps, defining the risk type, the spatial
relevance and a possible impact of climate change.
The risk typology focuses on the risk perspective instead of the hazard perspective. This
broadens the possibility of describing the interactions between hazards and the societal
reaction and response to hazards (for example aspects of risk perception). Both aspects
have to be considered in a risk management process. The German Advisory Council on
Global Change (WBGU) criteria served as a basis for the classification and
characterization of risks. This categorization of risks into certain types, however, does not
enable the extraction of those risks that are relevant to the ESPON context from the great
number of possible risks. For example, murder, drug abuse or road accidents definitely
belong to the highest risks in Europe, while a meteorite impact, although unlikely, could
destroy large parts, if not all, of Europe. But since these risks do not have a clear spatial
relationship, the selection of risks excludes certain risks by a spatial filter.
The spatial filter screens risks according to their spatial character. The occurrence of
spatially relevant hazards is limited to a certain disaster area that is regularly or irregularly
prone to hazards (e.g. river flooding, storm surges, volcanic eruptions). Spatially nonrelevant hazards can occur more or less anywhere (e.g. car accidents, meteorite impacts).
On the basis of these criteria, the hazards considered as of higher relevance for the EU
27+2 area in the ESPON context were selected. Of these, floods, flash floods, storm
surges, avalanches, landslides, droughts, forest fires, winter storms, and extreme
temperatures are assumed to be influenced by climate change.
Hazards are natural extreme events or technological accident phenomena that can lead to
threats and damages among the population, the environment and/or material assets. The
origin of hazards can be purely natural (e.g. earthquakes) or technological (e.g. accidents
in a chemical production plant), as well as a mixture of both (e.g. sinking of an oil tanker in
a winter storm and subsequent coastal pollution). Natural extreme events usually become
22
a hazard when human beings or material assets are threatened. All so-called natural
hazards occur on a more or less regular basis, as they are phenomena that belong to
natural processes. Being part of natural processes they do not pose any threat to the
natural system itself, as Nature is used to recover from natural hazards and adapt its life
forms to it. In extreme cases when humans influence natural hazards, e.g. arson in the
case of forest fires, these hazards are not purely natural any longer and can cause severe
damages to the nature itself.
Technological hazards pose threats to human assets and the nature, as they can have
impacts and cause pollution that are not natural processes. Also, technological hazards
can have very long lasting non-natural effects (e.g. oil spills and nuclear fallout). The focus
of the ESPON report lies on representing the natural and technological hazards in
administrative regions, on NUTS3 level (Nomenclature of Units for Territorial Statistic), of
the ESPON space. Since all of the EU 27+2 regions are populated and bear human
assets, all natural and technological phenomena that can be hazardous to human life,
properties, and the nature are defined as hazards.
All hazard maps follow the same classification of hazard intensity in five classes. In cases
where it was not possible to distinguish between five hazard classes, the same
classification range (very low: 1, to very high: 5) is kept, with fewer classes in between
very low and very high.
Whenever a multitude of hazards has to be considered in risk management, the question
of weighting the relevance of certain hazards appears. The Delphi method was adapted for
the specific use of hazard weighting and was tested several times in four case study areas
(the Dresden Region and the Ruhr District in Germany, The Centre Region of Portugal and
Regional council of Itä-Uusimaa in Finland) in the scope of the ESPON Hazards project.
To avoid distortion by regional bias, experts with a clear European perspective were
chosen; also the geographical provenance of experts was considered. Six of these experts
had southern European provenance and six represented central north European areas.
Results of the Delphi survey are represented by more or less calibrated average values of
the responses from the expert panel. During each round experts had the possibility to alter
the estimation after taking consideration of the average value from the previous round.
Ideally, the final estimation represents the so far final ‘opinion’ on the weight of each
hazard. The biggest emphasis was clearly on natural hazards (73.9 %) with floods (15.6
%), forest fires (11.4 %) and earthquakes (11.1 %) on the top of estimations.
Technological hazards in total received 26.1 % with major accidents hazards weighted
highest (8.4 %).
The aggregated hazard map (Figure 2.4(a)) shows that the highest hazard classes form a
kind of a scorpion-shape covering parts of southern, western, central and eastern Europe.
The two arms and the claws of this high hazard scorpion start off on the coastal areas of
the United Kingdom and the Iberian Peninsula, respectively, and the head is found in
central and southern Germany. The tail is then more scattered towards eastern Europe,
and finally turns southwards ending in Greece. Some hotspots are located outside of this
"high hazard scorpion", i.e. central Italy and parts of southern Scandinavia. Most of the
NUTS3 areas have a medium and some a low aggregated hazard. Besides scattered
spots, only few large areas have a very low aggregated hazard, mainly in northern Europe
and central-south France.
23
(a)
(b)
(c)
Figure 2.4 (a) aggregated hazard map (based on 15 hazard indicators); (b) Integrated vulnerability map; (c)
aggregated risk
In the map analysis one has to take into account that the 15 hazards used for these maps
are based on current knowledge that is comparable among all EU 27+2 countries. The
technological hazards are represented by only 4 hazard types. The maps thus serve as an
overview on the entire area, but regional and local analyses always have to take the best
available data into account.
Vulnerability and risk: In order to determine a risk factor, the ESPON Hazards project
acknowledges damage potential and coping capacity as the two main sides of regional
vulnerability. Within the ESPON context, Risk is defined as a “combination of the
probability (or frequency) of occurrence of a natural hazard and the extent of the
consequences of the impacts. A risk is a function of the exposure of assets and the
perception of potential impacts as perceived by a community or system. Risk = Hazard
potential x Vulnerability”. Vulnerability is understood as “the degree of fragility of a
(natural or socio-economic) community or a (natural or socio-economic) system towards
natural hazards”. It is a set of conditions and processes resulting from physical, social,
economic and environmental factors, which increase the susceptibility of the impact and
the consequences of natural hazards. Vulnerability is determined by the potential of a
natural hazard, the resulting risk and the potential to react to and/or to withstand it, i.e. its
adaptability, adaptive capacity and/or coping capacity. The overall regional vulnerability is
thus measured as a combination of damage potential and coping capacity. The following
formula is used: Vulnerability = Damage potential + Coping capacity, and for both damage
potential and coping capacity a set of indicators was chosen.
The following four chosen vulnerability components are hazard independent:
- The (high) regional GDP/capita measures the value of endangered physical
infrastructure and the extent of possible damage to the economy, according to
insurance company's point of view.
- Population density measures the amount of people in danger;
- The proportion of fragmented natural areas to all natural areas presumes that small
and fragmented areas are more vulnerable, since they are likely to be totally
destroyed if a hazard strikes;
24
-
The (low) national GDP/capita measures the capacity of people or regions to cope
with a catastrophe.
In the ESPON Hazards -project the national GDP/capita was used to indicate coping
capacity, since the presumption was that coping capacity is weak in poor countries and
strong in rich countries. It was further presumed that there are no marked differences in
coping capacity inside a country. Due to the fact that fragmented natural areas only refer
to a specific part of the ecological dimension, the indicator was only given the percentage
value of 10. The other three indicators thus received the value of 30%. Figure 2.5 shows
the integrated vulnerability index with the four feasible indicators. Using this information,
an integrated vulnerability map was created over the EU + 2 area.
Figure 2.5 Integrated vulnerability index
Combining the hazard and integrated vulnerability information, the aggregated risk map
(Figure 2.4(c)) was created. The final map is an integrated risk map that is a combination
of the vulnerability map (Figure 2.4(b)) and the aggregated hazard map (Figure 2.4(a)).
The risk maps follow a legend that displays the hazard values on the y-axis and the
integrated vulnerability described above on the x axis. All fields with the same sum (e.g. 4,
i.e. fields 3+1, 2+2 and 1+3) have the same risk towards a certain hazard. The different
shades of the same colour allow distinguishing between a higher intensity of a hazard or a
higher degree of vulnerability, respectively.
The aggregated risk map shows a similar pattern as the aggregated hazard map, even
though the scorpion shape of high risk has moved towards medium risk (classes 6, 7, and
8). The Pentagon Area displays the highest agglomeration of high risk, and the largest
parts with low risk are found in northern Europe.
Spatial patterns of hazards were studied by combining individual hazards on NUTS3 level.
The hazard interaction map is based on physical processes between hazards. In addition
to the development of the overall hazards interaction map, several hazard combinations
were studied on a European scale. In summary, the ESPON approach can be a valid tool
in order to define preliminary strategies and policies at a European level, however, the
limited series of homogeneous data available could under-estimate or over-estimate the
actual distribution of hazards and their potential impact on European countries.
25
2.2.4 The Applied multi-risk mapping of natural hazards for impact assessment
(ARMONIA) project (ARMONIA, 2007).
The overall aim of the research project ARMONIA (Applied multi Risk Mapping of Natural
Hazards for Impact Assessment) was to develop a new approach to producing integrated
multi-risk maps to achieve more effective spatial planning procedures in areas prone to
natural disasters in Europe. Amongst other things, the output of the ARMONIA project was
conceived to harmonize (1) the methodologies for hazard and risk assessment for different
types of potentially disastrous events and (2) the different processes of risk mapping in
order to standardize data collection, data analysis, monitoring, outputs and terminology in
a form useful to end users (multi-hazard risk assessment).
ARMONIA comprised the following steps: (1) analysis of the state-of-the-art for spatial
planning and mapping of risk from natural hazards; (2) development of a methodology for
harmonized integrated maps; (3) development of a harmonized knowledge base of
terminology; (4) integration of harmonized risk maps with spatial planning decision
processes in the form of a decision support framework; and 5) implementation, integration
and analysis of a case study simulation.
Theoretical approach of multi-risk analysis and mapping:
The main scope of the project was to produce a methodology that capably combines
multiple risks on a meaningful basis, taking into account all the basic and indispensable
parameters of risk analysis, such as hazard (where, when and how intense/severe a
natural event can be), exposure (typologies of elements at risk located in a hazardous
area), vulnerability (degree of potential damage expected by each relevant element at risk
vs. natural event occurrence), risk (degree of total damage due to the occurrence of one or
more different natural events in a given area).
Generically, risk assessment can define the level of expected damage for a given new
stock and then to define the required building regulation. In the case of existing goods we
consider the conventional formulation of risk (e.g. UN-ISDR):
R = H x EL
Where: EL = Va x V
and R = Risk, Expected losses (of lives, persons injured, property damaged and economic
activity disrupted) due to a particular hazard of intensity (magnitude) “I” for a given area
and reference period; H = Hazard, the threatening event, or the probability of occurrence
of an intensity (magnitude) “I” phenomenon, potentially damaging, within a given time
period and area (e.g. water level, discharge, stagnation); EL= Expected loss for the
threatening event, of an intensity (magnitude) “I”; Va = Value of exposed elements
(elements at risk); V = Vulnerability, degree of loss due to the occurrence of a particular
hazard of intensity “I” affecting an exposed element or a group of exposed elements, also
a-dimensional percentage of Va lost as consequence of a hazard of intensity (magnitude)
“I”.
In the project reports it is stated that multi-hazard (and multi-risk) cannot be the simple
combination (e.g. by superimposition of individual hazard maps and summation of hazard
degrees) of hazard categories together by assuming equivalence between a ‘high’ flood
hazard and a ‘high’ earthquake hazard. A feasible approach way may be combining
26
different risk estimates but only if a common and meaningful risk metric which works
between and across multiple forms of risk can be identified.
The best way to solve the problem has been envisaged in vulnerability analysis as key
element that links together the natural event, in terms of type and ‘dimension’, with the
exposed elements, both structural and non-structural. Recent perspectives on risk and
hazard management have emphasized the need to take into consideration distinct forms of
vulnerability, and in ARMONIA, a wide range of potentially exposed elements and
vulnerabilities was identified, and the ways in which some of these may be represented
through indices constructed from census and other spatial data in area, line and point form
were proposed.
One of the results achieved in ARMONIA is the definition of the possibility of mapping
multi-hazard/risk scenarios, considering the cumulative consequences (i.e. economic loss,
victims) of different natural events affecting the same exposed element. Theoretically, a
rigorous multi-risk mapping procedure is based on the following steps:
Step 1: identification of individual hazards for ARMONIA main spatial scales (strategic
regional, local general, local site);
Step 2: assessment of vulnerability functions for any individual category of natural event,
having as input the event location, intensity or severity parameters hazard category and as
output an average expected damage;
Step 3: assessment of fragility curves, when possible, for any individual category of
hazard, obtaining the probability of damage (e.g. for seismic hazard the % of cracks in
walls, the % of not statically safe buildings, the % of collapsing buildings) for given
categories of exposed elements defined by spatial planners;
Step 4: analysis of risk for any individual category of hazard;
Step 5: harmonization of different individual values of damage (risk), likely in terms of
fragility curves (probability of different damages for the same exposed element), for the
same return period.
Finally, for the realization of multi-risk maps, it becomes essential to construct vulnerability
curves having as input (x-axis) the individual hazard (e.g. intensity, magnitude, category)
and as output (y-axis) the average loss, possibly defined as probability of occurrence
(fragility curve). In fact, vulnerability can be expressed as the degree of loss to a given
element at risk, or a set of such elements (i.e. a system), resulting from the occurrence of
a natural or technological phenomenon of a given size. The induced aggravation (e.g.
caused by an earthquake, pollution, explosion, and landslide) should be expressed in
terms of a number of accessible and pertinent parameters when used for vulnerability
analysis. The choice of the most appropriate techniques to be used depends on the size of
the project area, the resources available and the data already collected. The reduction of
vulnerability may be studied in the general framework of a disaster resilience assessment
process. Vulnerability as well as resilience can be conceptualized along the four
interrelated dimensions:
-
Physical (e.g. destruction of a historical hotel);
Human, social and functional (e.g. life loss, employment, lodging);
Economical (e.g. reconstruction cost);
27
-
Identity-related issues (e.g. impact on the image of the town, tourism).
In practical terms, the main expected result is the construction of vulnerability functions,
possibly related with fragility functions, having a different x-axis for any natural event
typology and a common y-axis. Therefore, it would be possible to define all the risk with
the same factor. Figure 2.2.4-1 gives an example of the proposed vulnerability structure,
hereby considered as the correct conjunction point between hazard and risk; in a
preliminary stage the investigation could be restricted to the vulnerability function (Figure
2.6), instead of defining a comprehensive fragility scenario (Figure 2.7)
Figure 2.6 ARMONIA function between hazard types intensity (x-axis) vs. average damage of exposed elements
(y-axis) according to scale of analysis
Figure 2.7 ARMONIA function between hazard types intensity (x-axis) vs. probability of damage for different
categories of exposed elements (y-axis) according to scale of analysis
For the ARMONIA researchers, a multi-risk approach should be based on the production
of single and rigorous hazard maps. These maps have the main scope to depict all
considered risks to orient the spatial planning process and should contain the following
information in hazard analysis: (1) site or area of occurrence and potential development of
28
natural events; (2) intensity/severity/magnitude of the potentially disastrous event through
parametric scales; (3) return time of event, possibly related to potential triggering factors;
(4) hazard analysis should take into account minor, but more frequent events, as well as
major expected events characterized by highest intensity/lowest frequency since the latter
are the most potentially disastrous. On the other hand, it should be taken in account that
exposure and vulnerability factors are fundamental in risk analysis. These are often underevaluated or neglected in most current scientific and practical applications. The measure
of risk, that can be the expected loss or level of damage, can be represented in qualitative
or quantitative ways, depending on the spatial scales, such as:
-
costs to be spent for reconstruction;
damage index;
mixed quantitative/qualitative description of the expected damage, calculated by
means of probabilistic risk assessment and/or scenario approaches according to
spatial planning decisions and requirements.
Vulnerability and risk should be evaluated possibly in a deterministic/parametric manner or
qualitatively in case of systemic and organizational vulnerability. In any case, mapping
exposure and vulnerability can be considered as the minimum standard since complex or
complete models of vulnerability are actually inapplicable. Exposure and vulnerability
levels have to be referred to each typology of hazard in relation with the potential intensity
of events, especially at a local scale.
Following the investigation and the state of art analysis performed in the project, it was
defined that harmonization of natural hazard processes for land use planning and
management can be developed in the following way:
1. synthetic indicator of heuristic degree of multiple hazards affecting a given territory
(e.g. high, medium, low) or integer number of affecting natural hazards or simplified
multiple layers (hazards) of a summary map;
2. an integrated indicator of damage/losses, summing up, for a given time period (or
heuristically), multiple risks, individually evaluated;
3. a holistic approach of managing different hazards (e.g. to investigate
contemporaneously all the hazards affecting a given territory);
4. domino effects (e.g. landslide induced by earthquake).
Table 2.3 gives the potential spatial field of application for every identified procedure,
according to the state of art.
Regional
strategic
1
2
3
4
synthetic indicator of heuristic
degree of multiple hazards
integrated indicator of
damage/losses
holistic approach of
managing different hazards
(multi-layer)
domino effects
Local
General
Local
detail
X
X
X
X
X
X
X
X
Table 2.3 Potential spatial field of application (X mark) of suggested procedures following the state of the art on
multi-hazard risk assessment
29
Notes about the ‘draft’ ARMONIA methodology: practical application.
The ARMONIA methodology was applied in the Arno River basin in Italy, and at a national
level in England and Wales as a means to validate the methods developed as part of the
project. This exercise allowed the authors to identify a number of practical problems, gaps
and limitations of the ARMONIA methodology; in this way the ‘draft’ methodology is
outlined.
The key question that was required to be addressed as part of the ARMONIA methodology
is how to map risks from multiple natural hazards. Conceptually it is simple to map and
aggregate risk from a number of different natural hazards by assessing the annual
average risk for a particular risk metric for a particular hazard. However, for this approach
to be implemented the following need to be available: (1) The extent and intensity of the
hazard for a number of different return periods (i.e. annual probabilities) ranging from
“frequent” events (e.g. 1 in 2 year events) to “rare” events such as those with a 1 in 1,000
year return period; (2) Vulnerability functions for each hazard relating the intensity of the
hazard to the risk. It is often the case that where such functions are available they are
specific to geographical areas or countries. The theoretical method for assessing risk in
this manner is shown in Figure 2.8 (note that these are purely hypothetical vulnerability
curves that have been used to illustrate how assess the risk generated by natural hazards
can be quantified).
(a)
(b)
Figure 2.8 (a) Hypothetical vulnerability functions relating the intensity of a hazard to a risk metric; (b)
Aggregating risk from multi-hazards to assess the average annual risk (source: ARMONIA, 2007). Note: the
above curves are hypothetical probability versus risk metric curves to illustrate the method by which risk
metrics from different natural hazards can be combined
The ‘draft’ ARMONIA methodology looks to overcome the lack of detailed vulnerability
functions relating hazard intensity to a quantitative risk metric by using a variety of indices
to assess the risk for a particular natural hazard. These indices are as follows:
- Hazard indices;
- Consequence indices for different receptors or exposed elements (e.g. building,
people and infrastructure). It should be noted that the consequence indices
comprise a measure of vulnerability and exposure. These are combined to give a
consequence index.
30
The hazard and consequence indices are combined to produce a risk index for each
hazard. The draft ARMONIA methodology allows a weighting factor that must total 20 to
be applied to each consequence index by the relevant stakeholder. Figure 2.9 shows the
way in which risk is assessed using the draft ARMONIA methodology. It should be noted
that the draft ARMONIA methodology produces a combined risk score for each individual
natural hazard. It does not allow the risk from multiple natural hazards to be combined.
Figure 2.9 ARMONIA methodology for producing a risk index for a particular hazard
-Development of hazard indices: It is recommended that an approach that takes into
account both intensity and probability of the hazard is used. Such an approach is shown
diagrammatically in Figure 2.10. The intensity scales could be used in conjunction with
four return periods corresponding to a high, medium, low and very low probability. These
return periods could be as follows (for example):
-
<1 in 25 years
1 in 25 to 1 in 100 years
>1 in 100 years to 1 in 300 years
>1 in 300 years
High;
Medium;
Low;
Very low.
Figure 2.10 Diagram to show hazard levels and indices as a function of probability and intensity
31
Figure 2.10 provides four classes and indices of hazard. For example a medium hazard
with an index of 2 can result from a hazard with a low to medium intensity and a probability
of occurrence ranging from low to high. The draft ARMONIA methodology requires
hazards to be classified into non-dimensionalized indices. In practice hazards classified
into more than 10 categories become difficult to clearly display within a GIS environment.
The conversion of hazard data that represents a wide variety of units into integer values
may lead an end-user to interpret equal scores as representing an equal level of hazard
for different hazard types. For example, a flood with a “Class 2 hazard” should have the
same intensity and probability as a “Class 2 seismic hazard” if multi-hazard risk maps are
to be comparable.
Note that a more suitable universal method of classifying hazards would be to given the
annual probability of exceedance of a given hazard intensity. This is a useful measure
since it returns a value between 0 and 1 and is thus more readily compared across hazard
types.
-Development of Consequence indices: As part of the draft ARMONIA methodology,
consequence indices have been developed for a number of receptors (i.e. exposed
elements): (1) People; (2) Buildings; (3) Road network; (4) Agriculture; and (5) Other
buildings (e.g. commercial and industrial areas, airports).
Consequence indices have been developed at a regional level only for the following
hazards: Earthquakes, Floods, Landslides, Volcanoes, and Forest fires. The general
process for computing the consequence indices is shown in Figure 2.11.
Figure 2.11 Flow chart showing the method used for producing vulnerability and consequence indices
Finally, for the (multi) risk assessment, a Decision Support System (DSS) was developed:
The “Multi Risk Land Use Management Support System (MURLUMSS) DSS architecture”,
which encompasses the developed methodology. The DSS takes a consequence index
(where consequence = vulnerability x exposure) and average hazard index in each census
32
area and integrates them to produce a risk score between a theoretical minimum of zero
and maximum of 1. Figure 2.12 details the steps taken in the DSS to produce a risk score
for each census area. The DSS uses only one hazard per risk score but may use an
unlimited number of consequence indices. The weighting procedure allows stakeholder
participation through the subjective relative valuation of receptors within the risk score.
The risk score is normalized between the theoretical maximum and minimum possible
considering a census area under worst case conditions (highest hazard and consequence
indices). The minimum theoretical risk depends on whether all the census areas contain
receptors (i.e. exposed elements).
Figure 2.12 Methodology employed in the ARMONIA Decision Support System (DSS)
Risk analyses within the DSS are carried out through the implementation of a Multiple
Criteria Evaluation (MCE) ranking process. In the ranking process every criterion under
consideration is ranked in the order of the decision maker’s preference. To generate
criterion values for each evaluation unit, the risk for each receptor is weighted according to
its estimated significance. In the draft ARMONIA methodology the total weighting score
assigned to each risk index must total 20. The higher the weighting factor the more
prominence is given to the risk value for the receptor (For example, if three receptors (e.g.
people, buildings and roads) were being considered the end user may choose to weight
the risk factors for each of these as follows: people=12, buildings=5, and roads=3).
In conclusion, the draft ARMONIA methodology does not allow multi-risk maps to be
produced. The methodology does allow the consequences of a particular hazard on
33
different receptors (e.g. buildings, people) to be combined to produce a risk index for a
particular hazard. However, it is not clear whether such a risk index that combines a
number of social, economic and other risk metrics is of use to decision makers. On the
other hand, there is a high degree of subjectivity regarding the weighting and combining of
the vulnerability indices. There is no guidance in the methodology as to how these
weightings have been derived and how they affect the risk indicators. Finally, there is no
consideration of uncertainty in the draft ARMONIA methodology.
2.2.5 The Federal Emergency Management Agency initiatives: “Multi-hazard
identification and risk assessment” (FEMA, 1997) and “Understanding
your risks: identifying hazards and estimating losses” (FEMA, 2001).
The primary mission of the Federal Emergency Management Agency (FEMA) is to reduce
the risk of loss of life and property in the United States, and to protect U.S. institutions from
the disastrous effects of natural and technological hazards. FEMA accomplishes this
mission by leading, coordinating with, and supporting specialists at every level of
government (Federal, State and local) and the private sector in the development of a
comprehensive, risk-based emergency management program of mitigation, preparedness,
response, and recovery.
Identification of hazards and assessment of risks affecting the United States and its
territories are important steps in the process of reducing the impacts of disasters. These
steps help lay the foundation for the judicious allocation of finite resources to support
mitigation initiatives. HAZUS, The national risk assessment (loss estimation) methodology
under development by FEMA in cooperation with the National Institute for Building
Sciences, is intended to achieve this objective.
Based on the hazard identification and risk assessment research and evaluation
conducted for the “multi-hazard identification and risk assessment” report (FEMA, 1997),
the findings include:
-
-
-
-
Improvements are needed in the characterization of all hazards because there are
inconsistencies in the amount and quality of data available for each hazard;
Hazards must be better defined because of inconsistencies in definitions used by
Federal, State, and local government agencies and private-sector entities involved
in evaluating and mitigating hazards;
A model methodology for risk assessment for all hazards should be established,
and the level of sophistication associated with current methodologies should be
enhanced;
A more uniform technique to quantify numerically the risk of each hazard, on an
annual-percent-chance exceedance basis, should be developed to allow for a more
equitable comparison of risks for multiple hazards;
The results of risk assessments should serve as the basis for the prioritized
administration of mitigation programs and funding; and
Methods for evaluating the benefits and costs of mitigation programs should be
enhanced to include quantitative and qualitative elements.
The standard risk assessment (loss estimation) methodology developed jointly by FEMA
and the National Institute of Building Sciences (NIBS) is nationally applicable and
34
standardized. As originally developed, the methodology, referred to as Hazard United
States (HAZUS), is used to assess the risk of, and to estimate the potential losses from,
earthquakes. It incorporates the better features of previously developed loss estimation
methodologies and overcomes many shortcomings.
HAZUS is an integrated geographic information system designed for the personal
computer. It was developed based on several criteria: standardization; user-friendly design
and display; accommodation of user needs; accommodation of different levels of funding;
revisable results; state-of-the-art models and parameters; balance; flexibility in earthquake
ground shaking intensities; and non-proprietary methods and data. The HAZUS framework
includes six major modules: Potential Earth Science Hazard; Inventory; Direct Damage;
Induced Damage; Direct Losses; and Indirect Losses. The modules are interdependent:
the output from one module acts as input to another. The modular approach allows
estimates based on simplified models and limited inventory data. More refined estimates
based on more extensive inventory data and detailed analyses can be produced. Limited
studies can be conducted, which may be desirable because of budgetary and inventory
constraints. FEMA initiated development of HAZUS specifically for direct and indirect
economic and social losses from earthquakes and secondary hazards triggered by
earthquakes such as fires and floods due to dam or levee failure.
HAZUZ-MH is a nationally applicable standardized methodology and software program
that contains models for estimating potential losses from earthquakes, floods, and
hurricane winds. NIBS maintains committees of wind, flood, earthquake and software
experts to provide technical oversight and guidance to HAZUS-MH development. Loss
estimates produced by HAZUS-MH are based on current scientific and engineering
knowledge of the effects of hurricane winds, floods, and earthquakes. Estimating losses is
essential to decision-making at all levels of government, providing a basis for developing
mitigation plans and policies, emergency preparedness, and response and recovery
planning. HAZUS-MH uses state-of-the-art geographic information system (GIS) software
to map and display hazard data and the results of damage and economic loss estimates
for buildings and infrastructure. It also allows users to estimate the impacts of hurricane
winds, floods, and earthquakes on populations.
HAZUS-MH provides for three levels of analysis (Figure 2.13):
-
-
-
A Level 1 analysis yields a rough estimate based on the nationwide database and is
a good way to begin the risk assessment process and prioritizing high-risk
communities.
A Level 2 analysis requires the input of additional or refined data and hazard maps
that will produce more accurate risk and loss estimates. Assistance from local
emergency management personnel, city planners, GIS professionals, and others
may be necessary for this level of analysis.
A Level 3 analysis yields the most accurate estimate of loss and typically requires
the involvement of technical experts such as structural and geotechnical engineers
who can modify loss parameters based on to the specific conditions of a
community. This level analysis will allow users to supply their own techniques to
study special conditions such as dam breaks and tsunamis. Engineering and other
expertise is needed at this level.
35
Figure 2.13 HAZUZ-MH multi-hazard analysis levels
Three data input tools have been developed to support data collection. The Inventory
Collection Tool (InCAST) helps users collect and manage local building data for more
refined analyses than are possible with the national level data sets that come with HAZUS.
InCAST has expanded capabilities for multi-hazard data collection. HAZUS-MH includes
an enhanced Building Inventory Tool (BIT) that allows users to import building data and is
most useful when handling large datasets, such as tax assessor records. The Flood
Information Tool (FIT) helps users manipulate flood data into the format required by the
HAZUS flood model.
HAZUS-MH can perform multi-hazard analysis by providing access to the average
annualized loss and probabilistic results from the hurricane wind, flood, and earthquake
models and combining them to provide integrated multi-hazard reports and graphs.
HAZUS-MH also contains a third-party model integration capability that provides access
and operational capability to a wide range of natural, man-made, and technological hazard
models (nuclear and conventional blast, radiological, chemical, and biological) that will
supplement the natural hazard loss estimation capability (hurricane wind, flood, and
earthquake) in HAZUS-MH.
Furthermore, FEMA has also developed a companion tool called HAZUS-MH Risk
Assessment Tool (RAT) to help you produce your risk assessment outputs for
earthquakes, floods, and hurricanes. This tool was developed as a third-party model to
support HAZUS-MH and is used to display the outputs from the HAZUS-MH risk
assessment in an easy-to-use format. The RAT pulls natural hazard data, inventory data,
and loss estimate data into pre-formatted summary tables and text. These summaries can
support the presentation of data to decision-makers and other stakeholders and in the
mitigation plan.
Among the bibliographic material produced by FEMA, different “how-to” guides have been
produced, covering different topics as (1) getting started with the mitigation planning
process including important considerations for how you can organize to develop a plan; (2)
identifying hazards and assessing losses to your community and state; (3) setting
mitigation priorities and goals for your community; (4) evaluating potential mitigation
measures through the use of benefit-cost analysis and other techniques; (5) creating a
mitigation plan and implementation strategy; (6) implementing the mitigation plan including
project funding and revising the plan periodically as changes in the community occur; and
36
(7) incorporating special circumstances in hazard mitigation planning for historic
structures, among other topics (e.g. FEMA, 2001).
The FEMA guides (e.g. FEMA, 2001) are designed to provide the type of information US
states and communities need to initiate and maintain a planning process that will result in
safer communities. These guides have been carefully prepared so as to be applicable to
states and communities of various sizes and varying ranges of financial and technical
resources. They describe the creation of a composite multi-hazard map. Such a map
showing a composite of the areas of highest loss may be produced in various ways. The
map may indicate areas affected by multiple hazards as high loss potential areas and the
areas with one or no hazards as moderate or low potential loss. Another mapping
alternative is to identify areas with multiple critical facilities, major employers, repetitively
damaged structures, and infrastructure as high potential loss areas.
The FEMA report is intended to be an easy-to-understand guide for the field practitioner.
There is a clear audience for such a guide, and this report on identifying hazards and
estimating losses meets its stated objective in being simply and attractively written, with
colourful diagrams and in-set boxes to concentrate the attention of readers on the key
concepts needed for risk understanding.
As a document designed for ready
communication to decision-makers, it is commendable.
The FEMA “how-to” guides note the importance of compiling the results of work into a
written report to be presented to citizens and elected officials. The results of a risk
assessment will likely draw interest from a wide range of sectors in the community or state.
Business owners and residents will want to know what the results of the risk assessment
mean for them and what to do next. The FEMA report stresses the opportunity to use the
results of the risk assessment as a tool to galvanize the community and to secure interest
and support for the remainder of the hazard mitigation planning process. The risk
assessment can thus be an effective tool for public education, disaster response and
recovery, and economic development.
Applications of the various steps in the risk assessment process are illustrated through the
invention of a fictional community, the Town of “Hazardville”. Located in the imaginary
State of Emergency, Hazardville is developing a natural hazard mitigation plan, which
includes an estimate of potential losses. Hazardville is a small community with limited
resources and multiple hazards. The town has just established its mitigation planning
team, the Town of Hazardville Organization for Risk Reduction (THORR). A novel
entertaining way of putting across key messages is via a fictional periodical, ‘The
Hazardville Post’.
The FEMA (2001) report concludes with the statement that the loss estimate is the
foundation upon which a state or local mitigation plan is developed. With it, an official
should be able to identify what areas of your community or state are susceptible to each
hazard, where the highest losses would occur, how much a hazard may cost were it to
occur, and how the lives and quality of life in a community or state might be affected in the
aftermath of a disaster. This establishes a factual basis for developing a mitigation
strategy for a community or state, and constitutes important data necessary to support
future mitigation decisions.
37
2.2.6 The natural- and conflict-related hazards in Asia-Pacific project (OCHA,
2009)
This report presents the results of a study that aimed to quantify the risk posed by
earthquake, flood (and storm surge), landslide, cyclone and tropical storm, tsunami,
drought, and social unrest in form of intrastate armed conflict in the Asia-Pacific countries.
The study was commissioned by the United Nations Office for Coordination of
Humanitarian Affairs, OCHA Regional Office for Asia and the Pacific, Bangkok; and
financed by the Norwegian Ministry of Foreign Affairs.
A semi-quantitative approach adopted for the risk assessment was the same for all
countries in the study area to ensure consistent modelling for the whole region. The main
outcomes of the study were:
-
Hazard maps that show the geographical hotspot areas for single hazards and for
multi-hazards on a regional (sub-national) level;
Estimates of the exposed population to selected hazards;
Tabulated indices that attempt to rank the countries in terms of risk level; and
Recommendations for preventive actions in terms of how vulnerability can be
reduced and/or how coping capacity could be improved.
In the development of hazard maps, each natural hazard was classified into four
categories in term of its severity: negligible, low, medium and high. A key challenge in
the study was the quantification of the coping capacity of different countries in dealing
with these risks. Quantification of the coping capacity was essential in designing the
appropriate risk mitigation strategies.
In many situations there is interaction and strong correlation between the risks posed
by natural hazards and armed conflict. This interaction, however, was not evaluated in
the study. During the course of the study it became apparent that with the present
level of knowledge and availability of high quality data (or lack thereof), it is not
possible to estimate the risk in terms of expected mortality for all hazards. Therefore,
the exposed population was used as proxy for risk. The estimation of the exposed
population was based on a weighted average of the people living in areas with different
hazard severity categories, combined with spatial characteristics of the hazard in
question.
Methodology:
Each of the natural hazards considered in this project, as well as the conflict hazard, is
assessed in a different manner. This cannot be avoided because the spatial and
temporal scales and the frequency of occurrence vary over several orders of
magnitude for the different hazards considered. However, it does make it a challenge
to compare the risk associated with the different hazards, and to compare the risk
profiles of countries that are exposed to different types of hazard. The authors warn
that the procedure outlined in the project will probably not work very well for “very low
probability – extremely high consequence” events.
Hazard mapping: For each of the hazards considered in the study, 4 classes or
categories were defined on the basis of the computed hazard intensity and frequency,
or on the basis of an estimated hazard category:
38
0: Non-existent or negligible (white)
1: Low (green)
2: Medium (amber)
3: High (red)
For some of the natural hazards all the information is condensed into a single index.
For example for earthquakes, the Modified Mercalli Intensity (MMI) with a return period
of 475 years was used. For these hazards, it is relatively straightforward to classify the
hazard into 4 categories. For others, for example river flood and tropical cyclone, two
parameters (intensity and frequency) were used to define the hazard.
Calculation of exposed population and Risk Index: For each country in the study area,
the equivalent population exposed to each hazard was defined as 100% of the population
living in Hazard Category 3, plus 30% of the population living in Hazard Category 2, plus
10% of the population living in Hazard Category 1. For landslides, which have a limited
spatial extent even within a pixel of 30 arc_sec ×30 arc_sec, and for civil conflict, which is
impossible to resolve spatially to the same resolution as natural hazards, a correction
factor of 0.10 was applied to the equivalent exposed population.
An attempt was made to develop a global risk index for ranking the countries in the study
area. In general, it is logical for the risk index to have the following format:
Risk Index = function of
Coping Capacity Index
where
are weighting factors that designate the relative importance of different hazards,
is the equivalent population exposed to hazard “ ”,
is the ratio of
to the total
population,
is the number of fatalities caused by hazard “ ” within a reference time
frame, and
, which is a vulnerability indicator, is some function of the number of
fatalities divided by the equivalent exposed population.
The Coping Capacity Index for a given country is different for the different hazards.
However, in the project t was not possible to gather enough data in this study to come up
with meaningful coping capacity indices for different hazards and the aggregate index for
all hazards was used in the calculations.
The following equation for Risk Index was used in the study:
Risk Index =
Coping Capacity Index
The seven hazards considered in the calculation of the risk index were river flood,
earthquake, tropical cyclone (including storm surge), drought, precipitation-induced
landslide, tsunami and civil conflict. The following points should be noted about the
calculated risk index:
-
The fatality data for different natural hazards were obtained from the EM-DAT
database for the time period 1980-2007.
Fatalities due to civil conflict were based on PRIO Battle Death data 1980-2005 and
UCDP Battle Death data 2006-07. Battle death data for Pakistan are not available
and were roughly guessed for the calculations.
39
-
The weighting factors wi in the equation for the Risk Index are specified by the user.
The aggregate Coping Capacity Index varies from 2.06 to 4.83 (higher values
indicating higher coping capacity). These values were rescaled for use in the Risk
Index equation as follows:
Rescaled Coping Capacity Index =
Coping Capacity Index
-
The value of
-
The factor “1000” used in the Risk Index equation is purely for scaling purposes.
The values of exposed population to tsunami hazard are based on the population
data from the Year 2000.
was reset equal to 0.02 if it exceeded 0.02
The calculation of the Risk Index involves assigning weighting factors (importance factors)
by the user to the different hazards; the index, directly or indirectly, accounts for percent of
population exposed to different natural hazards and civil conflict, vulnerability, and coping
capacity. Table 2.4 is an example of the values of the Risk Index computed for the
countries in the study area using a specific set of weighting factors (i.e. with same
weighting factor for all hazards). The computed Risk Index is not very stable for the island
nations of the Pacific because of their small size and population. These island nations
should not be directly compared with the other nations in the study area.
2.3
Regional’ to ‘Local’ scale approaches
2.3.1 The “Natural Risk Assessment” (NARAS) project (e.g. Marzocchi et al.,
2009)
This project aims at contributing to the improvement and diffusion of these innovative
methodologies by:
- Disseminating and discussing the results obtained by EU funded projects dealing
with risk problems;
- Promoting the development of early warning methods for seismic risk mitigation;
- Stimulating the development of quantitative probabilistic methodologies of
evaluation of risks and different emergency scenarios using improved stochastic
methods;
- Promoting information and education actions in the schools aimed at increasing
consciousness of natural risks among young people;
In the pursuit of these objectives,there have been organized small workshops and
meetings wherein the possible actions to be implemented were discussed in depth and the
results reached by the relevant EU projects funded in the framework of FP5 and FP6 were
reviewed. The main outcomes of the NaRAs activities have been summarized in three
books, which deal with the state of the art of Seismic Early Warning, the implementation of
multi-risk approach, and seismology teaching at schools respectively.
40
Country
Bangladesh
Philippines
Indonesia
Myanmar
Nepal
Papua New Guinea
Japan
Pakistan
Bhutan
Sri Lanka
Malaysia
New Zealand
Viet Nam
Dem People's Rep of Korea
Cambodia
Thailand
India
China
Timor-Leste
Lao People's Democratic
Republic
Brunei Darussalam
Republic of Korea
Australia
Maldives
Singapore
Mongolia
Island nations of the Pacific
Micronesia (Federated States
of)
Vanuatu
Solomon islands
Samoa
Palau
Tonga
Fiji
Nauru
Kiribati
Tuvalu
Country Risk
Index
9.7
9.3
7.2
6.4
6.2
5.3
5.2
5.1
3.8
3.6
3.5
3.5
3.1
2.6
2.5
2.2
2.1
2.0
1.9
1.9
1.8
1.7
1.2
1.2
0.9
0.8
Index
10.7
7.8
6.0
4.7
4.5
4.4
4.2
3.2
2.3
2.3
Table 2.4 Risk index with the same weighting factor for all hazards
In a report on the multi-risk approach (Marzocchi, 2009), a quantitative procedure for
multi-risk assessment is presented, intended to make easier the comparison among
different threats and to account for possible triggering effects. In the document only the
major threats typical for Southern Europe were considered, which were the objectives of
the NARAS Project. Forest fires, snow avalanches, wind storms, heat waves are not
specifically considered, although the general methodology proposed can be applied to
evaluate risk related to these adverse events.
The document is divided in three parts: in the first part, the principles and rationales that
underlie the presented procedure for multi-risk assessment are explained; in part 2, a short
41
description of the most advanced procedures generally adopted to estimate individually
natural and anthropogenic risks representing major threats for Southern Europe is
provided; and in part 3, they tackle directly the problem of multi-risk assessment applying
innovative procedures and protocols to the case study of Casalnuovo , a town close to
Naples (Italy).
The multi-risk problem is split in two distinct phases: in a first phase, the whole set of risks
is homogenized to facilitate their comparison ranking. The common unit for comparing the
hazards is the risk of having a number of casualties; in the second phase, they explore in
detail possible “triggering” effects, showing how they can increase significantly the risk in a
specific site. In practice, cascading events are illustrated in a scenario approach: one
hazard creates an amplification of the risk for another hazard and changes the final risk
evaluation result.
Procedure for multi-risk assessment:
The purpose of multi-risk analyses is basically to establish a ranking of the different types
of risk taking into account possible cascade effects i.e. the situation for which an adverse
event triggers one or more sequential events (synergistic event). The procedure for
environmental risks assessment that includes the risk of multi-hazards synergy is
illustrated in this paragraph by identifying the main steps to be followed to estimate the
multi-risk index.
The main steps are those usually utilized to perform an environmental risk assessment
with some peculiar differences:
1. the estimate of the multi-risk index has to take into account possible cascade and/or
triggered related adverse events;
2. A common time frame and area under threat must be used;
3. A reference expected damage has to be defined “a priori”.
The general procedure for multi-hazard assessment is reported in the following scheme.
One peculiar aspect of this procedure is the creation of a set of scenarios correlating
adverse events from different sources. For each “risk scenario”, adverse events,
phenomena and damage will be correlated in a series-parallel sequence of happenings
through an “event-tree”. Each branch of the event tree will be quantified by a probabilistic
analysis of the “history” of the events, the vulnerability and the exposed values of the
specified targets. At last a final risk will be estimated. The procedure is summarized as
follows:
1. Identification of hazards/risk sources:
1.1. Risk sources identification (nature and location).
1.2. Characterization of adverse events and its propagation path.
1.3. Definition of possible single and multi-hazard scenarios starting by a given top
event and evaluating the possible triggering of other events.
2. Exposure and Vulnerability analysis:
2.1. Definition of exposure.
2.2. Phenomenon intensity distribution (e.g. ground acceleration, pressure waves,
distribution of chemical substance concentration for various areas, thermal flow,
etc.).
42
2.3. Identification of vulnerable elements (population at risk, strategic infrastructures,
historical structures, buildings).
3. Risk estimate:
3.1. Definition of the type of damage (e.g. reversible/irreversible damage to humans;
lethality; reversible/irreversible damage to the environment, damages to structures,
infrastructures, lifelines, economic damages, etc.).
3.2. Estimate of the entity of damage.
3.3. Probabilistic estimate of risk of each adverse event and of multi-risk.
3.4. Comparison between the multi-risk value and the “acceptable risk”.
For the multi-risk comparison it is useful the identification of a common reference damage
for all the single risks, for instance the risk of having a number of casualties. In fact once
the kind and intensity of the reference damage has been selected, different risks can be
ranked on the basis of their probability to originate the reference damage. This may
overcome the problem of assigning a monetary value to human life. This is needed if we
want to compare risks to damage structures or infrastructures with risks for human life.
The sequence of proposed steps is preceded by the characterization of the investigated
area and by fixing the time interval of reference: the extent of the area is defined case by
case, since the nature of the surrounding areas (type and number of vulnerable territorial
and environmental elements) and the extension of the consequences due to the events
may induce to expand or reduce the investigated area. The referring time interval is
chosen depending on the final goal of the risk analysis; for instance, the time interval can
be set to decades for land use planning, or few hours/days to manage an on-going
emergency.
Multi-risk assessment of a case-study: Casalnuovo (Italy)
The case-study of the Casalnuovo municipality, developed during the NARAS project, is
used to describe a possible procedure for multi-risk assessment. The Table 2.5 presents
the procedure used for single risk assessment.
Ranking the risks: The ranking of risks requires that each typology is calculated using the
same boundary conditions. First of all it is necessary to define a common timeframe, and
the specific kind of damage we are interested in. For the sake of example, in this report
they set the timeframe to one year and focus the risk analysis on a damage consisting of
human life loss.
After the specific analysis and calculations, the annual risk for human life is ranked as
follows:
As a result, it is stated that the volcanic risk is much higher than others, due to the
proximity of the Vesuvius volcano (about 12 km).
43
Hazard
Earthquakes
Hazard assessment
Probabilistic Seismic Hazard
Analysis from the INGV. The value
used is the PGA having 10%
chance of being exceeded in 50
years.
River floods
Landslide
Volcanic risk
Map from the River Basin
Authority
Bayesian event trees based on
Marzocchi et al. research works
(2004, 2008).
The ash load is the only volcanic
phenomenon considered for
Casalnuovo
industrial
An event tree methodology is
applied to a LPG storage plant. It
may lead to several phenomena,
among which jet fires.
Vulnerability
 The damage state (DS)
of buildings is evaluated
using fragility curves
corresponding to local
typologies.
 There is no causality in
DS0, D1, D2 or DS3
buildings. 0.5% of the
occupants are dead in
DS4 or DS5 buildings.
Statistical evaluation based on
historical data.
 Probability of collapse is
a function of ash load.
 The number of casualties
in collapsed buildings is
evaluated similarly to the
evaluation in the DS4/5
buildings after an
earthquake.
 Death probabilities are
associated with the
intensity measure
characterizing the
phenomenon. Thermal
flow is used for jet fires.
(e.g. death probability
reaches 1 when a person
is exposed to a thermal
flow over 12.5 kW/m2)

environmental Event tree for a release of
leachate from a landfill. 4 toxic
substances are considered and
the can impact health through
contaminated water ingestion.
The population exposed
are the people working in
the area
Slope Factor in mg/(kg.day):
dose to which a person can be
exposed daily, per body weight,
for each of the 4 substances.
Table 2.5 Procedures used for single-risk assessment (hazard and vulnerability approaches) in the Casalnuovo
case-study of NARAS project
Multi-risk due to triggering effects: The following scenario approach is used:
Volcano (risk source) → Eruption (event) → Ash fall (phenomenon) → building collapse
(damage) → Structural failure of civil infrastructures in the industrial plant (damage that
allows the “activation” of the risk source “LPG plant”) → Toxic release / fire / explosion
(event) → Substance leakage / Thermal Flow / Pressure Wave (phenomenon) →
Environmental air, soil, water contamination (damage that becomes risk source if exposure
pathways are active) → contaminant migration (event) → absorption by plants roots and/or
inlet into superior organism by direct exposition routes (ingestion, inhalation, dermal
44
contact) (phenomenon) → casualties, acute health damage, chronic health damage
(damage).
From this cascading sequence of events, it is concluded that the 10 cm ash load does not
threaten human life in a single risk approach, but it has an effect once the cascading
events are considered, as it increases the pipe-bridge probability of collapse, which leads
to casualties once the change is propagated down through the event tree.
In conclusion, in a single risk approach, this study considers a large number of hazards,
both natural and technological, and compares the associated risks quantitatively, using a
common damage unit (the probable number of casualties). This metric helps in the
homogenization of the results, however, a limitation of the approach used may be outlined,
since, for example, the death estimation in building collapsing under ash load is similar to
the one of buildings collapsing under earthquake ground motions, whereas this may be
different in terms of duration and potential pre-event evacuation.
The use of a multi-risk approach (which considers the possible cascading events)
accounts for scenarios that are not seen in a single risk methodology and leads to different
loss assessment results.
2.3.2 The “Risque Naturel Transverse” (RISK-NAT) project (e.g., Carnec et al.,
2005; Douglas 2005, Douglas 2007)
These documents discuss multi-risk assessments in general based on a literature review
and no case studies are presented. It is noted that hazard assessment for various natural
perils have considerable similarities in terms of, for example, the use of hazard maps for a
given return period and the separation of the evaluation into the assessment of ‘event’
parameters characterizing the event in general (e.g. earthquake magnitude) and ‘site’
parameters characterizing the impact of the event at a given point (e.g. peak ground
acceleration).
In the Douglas (2007) article, an interesting discussion about the use of fragility curves is
performed; it contrasts the situation for earthquake risk evaluation (and to some extent,
hurricanes) where fragility curves, to predict the damage level given a certain level of ‘site’
parameter, are commonly employed with other types of hazardous events (e.g. landslides)
where fragility curves are uncommon and rarely used. It is noted that this means that
quantitative risk evaluations are rarely made for these types of perils, which impedes the
study of the relative contributions from different hazards to the overall risk at a site.
It is argued that physical vulnerability is poorly modelled for these perils for many reasons:
the cause of human casualties (from the event itself rather than by building damage); lack
of observational data on the hazard, the elements at risk and the induced damage; the
complexity of the structural damage mechanisms; the temporal and geographical scales;
and the ability to modify the hazard level. This encourages efforts to be made to follow
examples for earthquake risk evaluation to improve risk (and not just hazard) evaluations
for other phenomena.
2.3.3 The comparative multi-risk assessments for the city of Cologne, Germany
(e.g. Grunthal et al., 2006; Kleist et al., 2006; Merz and Thieken, 2009)
45
Three different types of natural hazards (wind storms, river floods and earthquakes) are
quantitatively compared in terms of direct economic losses for the city of Cologne,
Germany. The aleatory variability of the natural hazards is taken into account by
evaluating hazard levels corresponding to annual probability of exceedance. The final
result is the direct economic losses against the annual probability as presented in Figure
2.14.
Figure 2.14 Final output of the multi-risk analysis done for the city of Cologne, Germany (Grünthal et al., 2006):
Risk curves of the hazards due to windstorms, floods and earthquakes for the city of Cologne for losses
concerning buildings and contents in the sectors private housing, commerce and industry (reference year:
2000).
The approach follows the following steps:
1. Hazard assessment;
2. Asset inventory (common for all three hazards);
3. Vulnerability assessment.
This then leads to:
4. The loss estimation;
5. The quantitative risk comparison (final result)
Each of these steps may be divided in sub-tasks, as presented below:
46
2 – Asset inventory
1 – Hazard assessment
-
Choice of a intensity measure i1 for
which data is available
Collection of historical data for i1
Choice of an extrapolation
distribution law
Exceedance
probability
Relating i1 to another intensity
measure i2 which is more related to the
losses
-
Distinction of several land uses
Spatial use mapping
€/m2 value for each sector
based on the fixed assets value,
discounted in 2000 €s.
 Map of the land uses and
reconstruction cost per kind of land
use
Hist. Data
Extrapolation
i1
3 – Vulnerability assessment
-
Law relating the expected
losses to the intensity measure
i2
 Estimation of the damage ratio
Figure 2.15 Sub-tasks for each step of the risk assessment process
Hazard assessment
The methodology implies the choice of an intensity measure for which historical data is
available. Large events are interesting from a risk management point of view but may not
have occurred during the observation period. Distribution functions are therefore used to
extrapolate the limited data to the long return time periods. An expert has to choose the
distribution function as different ones may show the same good agreement with the data
but lead to very different extreme values.
Another intensity measure , which is better correlated to the losses, may also be
introduced and related to . The parameters presented on Table 2.6 are used for each
three hazards.
Asset inventory
Several land uses (e.g. agricultural, private housing …) are distinguished as they have
different economical values. They also have different vulnerabilities to floods. Among
these sectors, three categories are considered for the direct economic loss estimation:
private housing, commerce and industry. It represents 33% of the surface and 83% of the
value of the city. Assets of residential buildings are provided by Kleist et al. (2006).
47
Hazard
Intensity measure i1 : available in the historical
data set
i1
Historical data
Extrapolation
set
law
Windstorm
Mean
wind
velocity
River
floods
Annual
Maximum
Series :
annual
maximum
flow rate
of the
Rhine
PGA?
Earthquake
Hourly wind
speeds, 10 min
averaged
1971-2000
Daily gauge
from 1880 to
1999
Gumbel type I
Pearson type
III
Not detailed, based on Grünthal
& Wahlröm, 2006. Different
combinations of 14 seismic
sources zones are used. Logic
tree technique.
Intensity measure i2 relevant for the loss
estimation
i2
Correlation
Spatial
between i1 and
variation
i2
Peak
Empirical model No
gusts (3s) using local
velocity
parameters
Water
level and
inundation
areas
Seismic
intensity
-river discharge
 water levels
- water levels 
inundation
areas : (Digital
Elevation
Models)
Consider a “soft
to stiff
sedimentary”
soil.
Yes
Negligible.
No local
site effect
Table 2.6 parameters used for the hazard assessment (Grunthal et al., 2006)
Vulnerability (Table 2.7)
Hazard
Windstorm
River floods
Earthquake
Methodology
Empirical damage function developed by Munich Re relating Loss
ratio to gust speed
- The inundation area is divided in 50mx50m cells, which are
supposed to belong to a unique economical sector.
- Damage ratio function depending on the economic sector and the
inundation depth are used.
- Field survey and vulnerability classification (EMS-98) of 800
buildings in a test area.
- Fragility functions from Raschke, 2003 (apparently).
Table 2.7 Summary of vulnerability assessment methodologies
The methodology described is intended for a local (city) scale. The Spatial resolutions
used are: Grid for the damage function to windstorms: 0.1°x0.1°; Digital elevation model
for floods: 50m grid size.
As an added value of the methodology, the loss estimation through a common loss unit
(chosen to be the direct economic losses) makes the risks associated to different natural
hazards quantitatively comparable. However, it should be underlined that whether the
choice of the direct loss is the most appropriate is questionable. In fact, the effect of
different hazards has different time characteristics. For instance, the recovery of
constructions is not the same of that of agricultural land or trees. Different return periods,
for different hazards, make it also difficult to integrate the cost over a given period of time.
Besides these aspects that need attention, the presented results are interesting.
In general it is preferable to cover the whole loss estimation process: probabilistic hazard
assessment, vulnerability assessment and loss estimation. That will ease a quantitatively
48
comparison of two different hazards, or even damages to buildings: earthquakes will
mostly create structural damages to the structures, whereas floods will first impact the
equipment and goods.
However, for some hazards, the evaluation of the vulnerability is not easy and in that case
it may be acceptable, as a first approach, to go directly from the hazard assessment to the
loss estimation if reliable and enough information from the past (similar) events does exist.
On the other hand, some limitations can be mentioned: first, as written, “the risk curves
were calculated only for the most probable estimates, i.e. uncertainties were not
considered.” (Grünthal et al., 2006); for instance, in Grünthal et al. (2006), the statistical
uncertainty of the hazard is considered by expressing it for a range of return time periods.
Other uncertainties (epistemic…) are presented for the seismic hazard and are mentioned
(choice of a distribution function) for floods and windstorms, but are not propagated after
the hazard step. In Merz and Thieken (2009), uncertainties for the flood risk are
considered and propagated though the loss estimation process (e.g. see Figure 2.16).
Figure 2.16 (from Merz & Thieken, 2009). The left figure is similar to one from Grünthal et al. ( 2006) for floods
with, in addition, a cluster of curves corresponding to the evaluation and propagation of uncertainties through
the loss estimation. The contribution of different steps is illustrated on the right figure.
Second, the three considered hazards are among the ones that can be assessed
statistically; it may be more difficult to include other types of hazards in this framework
because they are usually assed using scenario-based deterministic methods. Finally, only
potential direct and tangible damages are considered, which is only a part of the total
losses.
2.3.4 The regional-level multi-risk project in the Piedmont region, Italy: “A
methodological approach for the definition of multi-risk maps at regional
level: first application” (Carpignano et al., 2009)
The aim of the paper is to describe the methodology behind the production of regionalscale multi-risk maps. In the paper it is suggested that the shift from single to multi-risk
approaches is one where there is a change from a hazard-centred perspective (singletype) to a territorial one, meaning a piece of territory composed of elements at risk that are
vulnerable in different ways to various sources of hazard. They therefore divide the multirisk approach into multi-hazard and multi-vulnerability approaches.
49
The methodology was defined in accordance with the current literature and various pieces
of legislation with regards to natural and technological risk management, in particular the
“Guidelines for the editing of communal plans of civil protection – Piedmont Region”
(Piedmont Region, 2004).
In the paper there was little written about the methodologies used in defining the hazards,
referring simply to various Italian and European legislations.
The conceptual model for vulnerability and risk assessment followed is summarized by
Figure 2.17.
Figure 2.17 The vulnerability model proposed by Carpignano et al. (2009)
An initiating event, IE, triggers a sequence (scenario) which results in consequences, C,
over an area П. The potential damage, Dp, is for exposed elements, E, the exposure
being a function of territorial density. Vulnerability, V, converts Dp to real damage, Dm.
Vulnerability is a function of susceptibility, S, and coping capacity Cc.
Expressing the model displayed in Figure 2.17 mathematically considering different
sources of hazard (seismic, industrial, flood, landslide, forest fire) resulting in different
scenarios (earthquake, industrial explosion, industrial fire, industrial toxic release, flood,
landslide, forest fire) which causes a mitigated type of damage (e.g., deaths, injuries,
collapsed buildings etc.), with eight indicators of damage being defined. Considering i as
representing a scenario and j a type of damage gives mitigated damage DM,j,i. This leads
to an i × j matrix which is used to display the risk indices Rj,i (Figure 2.18).
50
Figure 2.18 How the combination of scenarios (i) and damage indicators (j) determine the risk indices Rj,i as
used by Carpignano et al. (2009).
Each hazard is assessed by a historical analysis of past events and of land characteristics
(e.g., presence of industries, rivers etc.). The natural hazards are done from a statistical
point of view, while the industrial hazards are from an analytical basis. These analyses
thus determine the likelihood of an event of intensity m in terms of frequency or probability
(m being dependent upon the hazard of concern). Hence risk Rj,i is expressed as:
where
represents the probability of scenario . This may be shown by Figure 2.18,
where the lists of scenarios and damages are given, as well as the risk matrix, with those
combination of scenario and damage that are not relevant being marked in grey.
The mitigated damage is derived from the potential damage, modified by the susceptibility
and coping capacity. The potential damage in turn is found by combining impact area
maps with the maps of exposed elements (target density maps). The impact area maps
are derived from various sources based on the requirements of Italian legislation which
results in hazard zoning for each of the hazards considered.
Vulnerability assessment is discussed at some length. Calling upon a scenario i, damage j
from an intensity m, two components are calculated, susceptibility (S j,i,m) and coping
capacity (Cc,j,i). The corresponding vulnerability (Vj,i,m) is given by:
where
is the optimal level of reduction of the original susceptibility, where in this
case, the Piedmont authorities consider the optimal level to be the total reduction of
susceptibility, hence,
. This gives a coping capacity equal to 1 (or 100%) or complete mitigation.
By contrast, a value of 0 means that there is no damage reduction.
51
Susceptibility in this work refers to the probability of damage imputing to intrinsic
characteristics of territory and its elements, e.g., land topography dictates what areas will
be flooded. Estimating is carried out using information from the literature and experience
from natural events and industrial accidents. This requires searching for correlations
between intensity of events and the percentage of potential damage (e.g., for seismic
risk, correlation of building susceptibility to intensity values and building categories).
Coping capacity as defined in this work deals with activities devoted to mitigating the
effects of accidents/natural disasters that guarantee the safety of people and assets. The
values of the coping capacity parameters (dealing with monitoring networks, instruments,
finances, etc.) were assessed in accordance with expert judgement (civil protection
technicians, based on a questionnaire). Civil protection factors, which are used to
evaluate coping capacity, were derived from the questionnaire and were grouped into civil
protection components which were (see the paper’s Table 3):
-
Monitoring system, defined by their characteristics (preventative, manual, etc.);
Alert system, defined by their characteristics (manual, instrumental, etc.), human
resources etc.;
Protection measures for population, plans, procedures, training etc.;
Protection measures for infrastructures, quality, procedures, training etc.;
Assistance measures, accommodation, sanitation etc.
The coping capacity was then determined as the sum of performance degrees of each
factor weighted by importance. It was then found with respect to a scenario and type of
damage as
comp
comp
where
comp
fact
fact
such that comp is the importance level of the civil protection component,
is the
comp
performance level of the civil protection component, fact is the importance of level of the
civil protection factor and fact is the performance level of the civil protection factor.
The cumulative risk is calculated by:
where represents the total risk for a type j of damage summed over all scenarios i, with
the total risk given by
where R is the aggregated risk index and is the weight attributed to j by stakeholders.
The results themselves are presented in the form of maps of
where the risk for each
municipality is expressed in terms of the number of standard deviations from the regional
average, showing what parts of the region are subject to greater risk than others for a
given type of damage.
In the application on the Piedmont Region, Italy, earthquakes, industrial activities (e.g.
chemical-industrial plants and transportation of dangerous goods), floods, landslides, and
52
forest fires were the hazards considered for the analysis. These are then investigated in
terms of seven scenarios (referred to as in the above discussion): seismic events,
explosions, fires, toxic release, floods, landslides, and forest fires. Figure 2.19 is an
example of an aggregated risk map.
Cascade effects are mentioned but not considered in the presented methodology, but it
does discuss a summation of the risk resulting from the different hazards considered.
However it is more a sum of single-risk assessments.
As added value, we can mention that the outline of the concepts of multi-risk is useful; the
description of how vulnerability may be broken down into other components is well
outlined. The main value is the provision of summing multiple single risks arising from a
variety of hazard scenarios and damage types.
Figure 2.19 Example of an aggregated risk map edited by the Civil Protection Department for delivery
infrastructures which are exposed to five hazard scenarios: seismic event, industrial accident, flood, landslide
and forest fire (Carpignano et al. 2009).
2.3.5 The regional-level initiative of integration of natural and technological
risks in Lombardy region, Italy (e.g. Lari et al., 2009)
The methodology, intended to be applied at a regional scale, integrates information with
different degrees of accuracy into an indicator based approach, in order to develop a
regional scale multi-risk assessment and to identify “hot spot” risk areas for more detailed
analysis. A relative physical risk indicator based on hazard, vulnerability and exposure was
developed with an aggravating factor used to account for societal coping capacity.
Weighting methods (based on Budgetary Allocation, Fuzzy Sets and Analytic Hierarchy
Process) were used at two levels, first at an individual hazard level to assess hazard,
vulnerability and exposure into a relative risk indicator and then to integrate risk for
53
different hazard types into a multi-risk indicator. Sensitivity analysis for varying weights
was carried out.
In the study, landslides, avalanches, floods, wildfires, seismic, meteorological (lightning),
industrial (explosion-related accidents) risks were considered; road accidents, and work
injuries were also discussed.
The paper provides a practical example illustrating some of the difficulties that often will
occur in terms of homogenization of methods for different hazard types: different traditions
among professionals, different data types/sources, missing data, different analysis
methodologies, comparing very different risk levels. The paper also treats central topics
like sensitivity analysis and the importance of risk perception in weighting processes.
2.3.6 A conceptual model for “Reconsidering the risk assessment concept:
Standardizing the impact description as a building block for vulnerability
assessment” (Hollenstein, 2005)
The article proposes a new concept of risk assessment in which hazard impacts and
vulnerability are standardized. As explained in the abstract, “this is achieved by defining
states of the target objects that depend on the impact and at the same time affect the
object’s performance characteristics. The possible state variables can be related to a
limited set of impact descriptors termed generic impact description interface. The concept
suggests that both hazard and vulnerability assessment models are developed according
to the specification of this interface, thus facilitating modularized risk assessments”.
Hazard standardization:
The hazard impact is characterized based on the intensity of m generic components
. A preliminary list of generic components is suggested (see Table 2.8):
Table 2.8 list of generic components that may characterize the hazard impact intensity (I)
Wrapper functions are used to convert the output of existing hazard assessment
procedures into the generic hazard components.
54
Vulnerability standardization
The performance of the target object is also characterized by a vector S of n performance
characteristics s evaluated between 0 and 1.
The vulnerability V of the target is expressed as the relative change in performance for a
generic impact (applicable to all objects) min
:
An intermediary “State” step is also introduced between hazard and performance in order
to reduce the chances the important failure mode would not be considered.
The goal of the methodology is to be applicable to all hazards. Wind, earthquakes and
floods are used as examples in order to present how they can fit into the standardized
hazard model framework.
As added value, the paper proposes a framework for standardizing the risk assessment
for all hazards. It is based on the decomposition of the hazard impact in several physical
effects (acceleration, force, pressure…) that are not hazard-specific.
This is still a conceptual model, which is a limitation; the model has to be applied and the
accuracy of the results has to be shown (acceptance issues).
2.3.7 The Cities project for geohazards in Australian urban communities (e.g.
Grander et al., 1999; Granger and Haine, 2001);
AGSO - Geoscience Australia is Australia’s leading national geoscience research
organisation. The AGSO Cities Project assesses the effects on 27 urban communities of a
range of natural hazards. The AGSO Cities Project was established in 1996 to undertake
research directed towards the mitigation of the risks faced by Australian urban
communities that are posed by a range of geohazards. Geohazards are broadly defined to
include all earth surface processes with the potential to cause loss or harm to the
community or the environment. The ultimate objective is to improve the safety of
communities, and consequently make them more sustainable and prosperous.
Their approach in developing a multi-hazard risk assessment, as depicted in Granger and
Jones (2001) was not really an attempt to deal with multi-type hazard or risk as is
understood within the MATRIX project, it was more the case of multiple single-type hazard
and risk. However, some discussion of the correlation of hazards was made (e.g.,
Granger, 1999).
The scale involved was mainly urban community, although some regional (Southeast
Queensland) were considered. The communities and areas considered were: Cairns,
Mackay, Gladstone, Newcastle, South-east Queensland (mainly Brisbane), and Perth.
Table 2.9 outlines the main hazards considered; note, not all hazards were treated for all
areas. Likewise, when considering storms, there were some additional subdivisions for
specific cases, but these have been combined here for simplicity (thunderstorms and other
severe storms). Other hazards were considered in the IID (2006) report, many involving
possible industrial or technological incidences, e.g., contamination and pollution, structural
55
failure and fire, transport accidents, epidemics (human, animal and plant) and rarer natural
hazards, such as meteorite impacts.
Cairns
Mackay
Gladstone
SE
Queensland
Tsunami
Earthquakes
X
Landslides
X
Cyclones
X
Floods
X
X
Newcastle
Perth
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Stormtides
X
X
Thunderstorms
and other
severe storms
X
X
X
X
Heatwaves
X
X
X
X
Bushfires
X
X
X
X
Droughts
X
Fog and frost
X
Subsidence
X
Coastal erosion
X
Table 2.9 Considered hazards in the Cities project for geohazards in Australian urban communities (e.g. Grander et al., 1999)
For example, in their report for the South-East Queensland case they have:
-
established the risk study context and process;
identified the key risks faced by the South-East Queensland community that are posed by a
range of natural hazards; and,
analysed and characterised those risks.
Risk evaluation and prioritization (example from the SE Queensland)
The method that has gained wide recognition amongst Australian emergency managers is
the ‘SMAUG’ approach (which stands for Seriousness, Manageability, Acceptability,
Urgency and Growth) based on the work of Kepler and Tregoe (1981). The method
involves rating each risk in relation to these criteria as being high, medium or low. The risk
management standard (AS/NZS 4360:1999) provides a similar approach based on a
matrix to rate risk likelihood qualitatively against its consequences.
Although both of these approaches provide a useful method for reaching a qualitative (and
subjective) evaluation of risk, especially for a single hazard impact on a relatively small
community, they are significantly less useful when applied to a multi-hazard risk evaluation
and prioritisation for very large and complex communities such as that covered in this
study. The quantitatively-based total risk approach that Granger and Jones (2001) have
adopted in this study is intended to provide a more objective means of identifying the risks
that pose the greatest threat to the South-East Queensland community and to its
constituent neighbourhoods. To achieve this they have measured the input of the three
key variables in the total risk relationship, namely the hazards, the community elements
exposed to those hazards and the relative vulnerability of those elements.
Risks compared: Although there is still some way to go before we can produce a single
statistic that is able to measure total risk across all hazards, the authors argue that they
56
have been able to produce statistics that do provide a comparison of the community’s
exposure to some major hazards.
Building damage is the single, best indicator of risk. That is, the level of building damage
can be used to rank risks from various natural hazards (when considered against its
probability of occurrence). Building damage can also be used to estimate risk in absolute
terms, although such estimates will be incomplete. Other potential direct and indirect costs
to the community, for example from casualties or from business interruption, are also
important sources of risk. However, the damage to buildings may be the largest
component of direct damage from natural hazard disasters.
The use of building damage as a surrogate indicator for exposure across all elements at
risk clearly imposes limits. However, it is clear that in Australian urban areas, damage to
buildings has been the largest component of both direct and indirect damage in natural
disasters.
A comparison of estimated residential building damage from earthquakes and from severe
winds from tropical cyclones, for the entire South-East Queensland region, is shown in
Figure 2.20. For earthquakes, damage losses are defined in terms of the percentage of the
repair or replacement cost of a ‘typical’ residence including contents or, alternatively, the
percentage loss of the repair or replacement cost of the entire residential building stock
and contents. For cyclonic winds, damage losses are expressed as a percentage of the
total insured value of a ‘typical’ residential building and its contents. Alternatively, the
damage losses for wind can be considered as a percentage of the total insured value of all
of the residential buildings and their contents. The damage losses are the minimum losses
expected for the probability associated with that loss. For example, for earthquakes, there
is an annual probability of 0.5% that damage losses will be at least 0.007% of the total
insured value.
The uncertainties in the results are not shown in Figure 2.20. Because of the very large
uncertainties, the risks posed by earthquakes and by severe winds from tropical cyclones,
as determined from residential building damage alone, cannot be distinguished apart in
Figure 2.20. Risk is related to the area underneath the curve. An important feature of the
risk from tropical cyclone winds, and especially from earthquakes, is that a significant
amount of all the risk from these hazards is attributed to extreme events. That is, relatively
frequent events will not cause high levels of damage, particularly from earthquakes.
However, very rare events have the potential to cause large amounts of damage (and
earthquakes apparently more than cyclonic winds).
57
Figure 2.20 Comparison of percent building and contents damage from earthquake and cyclone wind
The risks posed by storm tide inundation and flooding are also compared. Structure and
contents loss curves for flood damage (Blong, 2001) are used to estimate the percent
damage loss. The impact of storm tide inundation compared to flood inundation when
determining damage estimates is unclear. It is considered, however, that salt water will
inflict greater damage and this has been incorporated in the storm tide damage loss shown
in Figure 2.21. Though no modelling was available for riverine flood with AEPs less than
1%, damage from riverine flood significantly exceeds damage from storm surge inundation
for AEPs of 2% and 1% (Figure 2.21).
Although Figure 2.20 and Figure 2.21 are both based on percent damage loss they are not
directly comparable. Earthquake and tropical cyclone wind damage losses are based on
aggregated damage estimates from numerous scenario events randomly imposed on
areas within the region, whereas storm tide and flood damage estimates are based on
inundation levels with a particular AEP imposed over the entire region. It is considered that
although the latter estimates will overestimate regional risk, they represent the numbers
that disaster managers need to base their plans, especially for carrying out precautionary
evacuations. Clearly, the estimates of damage in the figures contain many limitations and
uncertainties, and the comparisons of risk must be taken as indicative.
58
Figure 2.21 Comparison of percent building and contents damage from storm tide inundation and flood
2.3.8 A multi-hazard risk assessment for the Turrialba city, Costa Rica (e.g., Van
Wasten et al., 2002);
In the framework of the UNESCO sponsored project on “Capacity Building for Natural
Disaster Reduction” a “Regional Action Programme for Central America” was established.
Within this project a number of case studies throughout Central America are carried out.
The first of these is the multi-hazard risk assessment of the city of Turrialba, located in the
central part of Costa Rica.
The objectives of this study were to support the local authorities with basic information
required for disaster management at the municipal level, through the development of a
GIS database, containing the following types of information:
-
-
-
Hazard maps indicating the probability of occurrence of potentially damaging
phenomena within a given time period. This was done by generating hazard maps
for earthquakes, flooding and landslides for different return periods;
A database of elements at risk, concentrating on the buildings and the infrastructure
in the city;
Analysis of vulnerability of the elements at risk, taking into account the intensities of
events as indicated in the hazard maps, combined with information from damage
curves;
Cost estimation of the elements at risk, concentrating on the buildings and their
contents;
Multi-hazard risk assessment.
59
The reference describes a GIS-based system for risk assessment. The work uses an
orthophoto as the basis on which all buildings, land parcels and roads, within the city and
its direct surroundings are digitized, resulting in a digital parcel map. Then a number of
hazard and vulnerability attributes for the city are collected in the field and from desk
studies. The cadastral database of the city is used in combination with the various hazard
maps for different return periods to generate vulnerability maps for the city. The
vulnerability maps are combined with the cost maps of the elements at risk and the hazard
maps per hazard type for the different return periods, in order to obtain graphs of
probability versus potential damage.
Hazard assessment:
In the study three types of hazards were analysed: seismic, flooding and landslide hazard.
Probabilistic methods were used in order to obtain the respective values of PGA (peak
ground acceleration) of rocks for different return periods, based on the work by Laporte et
al. (1994) and Climent (1997). After soil amplification correction, these PGA values were
converted to the Modified Mercalli Intensity using the relation from Trifunac and Brady
(1975); the analysis resulted in four MMI maps for return periods 25, 50,100 and 200
years. Flood hazard maps were made related to two different phenomena: lateral erosion
hazard and inundation depth. As indicated before, flood depth maps were made using
historical information from field questionnaires. The resulting point file was converted into a
raster map in GIS using contour interpolation and point interpolation. The resulting flood
depth maps, for return periods of 25, 50 and 75 years, were classified into a number of
classes. Landslide hazard was determined based on the historical landslide inventory and
a number of factor maps, using a statistical approach.
Vulnerability Assessment:
In this study vulnerability assessment was only carried out for the buildings and the
contents of buildings, and basically only looking at direct tangible losses. The basic
method used was the application of damage-state curves, also called loss functions or
vulnerability curves (Smith, 1994). The cadastral database of the city was used, in
combination with the various hazard maps for different return periods to generate
vulnerability maps for the city.
Damage due to flooding depends on several factors, such as water -depth, duration of
flooding, flow velocity, sediment concentration and pollution. The study only focused on
damage related to water depth, and to velocity in the case of lateral erosion. In the report it
is argued that flooding time is generally not very important since most events are related to
flash floods with limited duration. The method used in this study for flood vulnerability
assessment can be considered as a GIS-based hybrid between the actual flood damage
approach and the existing database approach. This is because the vulnerability
assessment is based on a detailed database of elements at risk and on field data
collection related to the 1996 flood reported damages. Depending on the building type and
the number of floors a degree of loss (ranging from 0 to 1) was assigned to each category
of elements at risk, in relation with the different floodwater depth classes used. Separate
values were assigned for the expected losses related to the contents of buildings. In the
case of lateral erosion vulnerability was assumed to be 1 (complete destruction) both for
the building as well as for the contents.
For the determination of seismic vulnerability, the MMI maps were used in combination
with vulnerability functions for different types of constructions adapted from Sauter and
60
Shah (1978), who elaborated functions for Costa Rica as a whole. Vulnerability
assessment of population for seismic events was made according to the Radius method,
based on the building vulnerability and the type of building (residential, school, office etc.)
assuming two different scenarios: during daytime and night-time.
For the landslide vulnerability the size of the potential landslide area determined whether
the vulnerability was 0, 0.5 or 1. All vulnerability data was used in GIS to generate
vulnerability maps for each type of hazard and return period.
Cost Estimation (value):
In order to determine the value of the elements at risk, differentiation was made between
the costs of the constructions and the costs of the contents of the buildings. The costs of
the buildings were determined using information from real estate agents and architects in
the area. A cost per square meter was entered in the attribute table linked to the cadastral
map, and the cost per parcel was obtained by multiplication with the area of the parcel,
and the number of floors. A correction factor was applied related to the percentage of the
plot, which was actually built-up area, and also a depreciation factor was applied related to
the age of the buildings. An estimation of the cost of the contents of buildings was made
based on a number of sample investigations for different building types and socioeconomic
classes within the city. Based on the cost information three raster maps were generated:
one representing the building costs, one representing the construction costs, and one for
the total costs
.
Risk Assessment:
In this project, risk is understood as the expected degree of loss due to potentially
damaging phenomena within a given time. In this case there are many different potentially
damaging phenomena with different return periods. Therefore risk was determined by first
calculating specific risk for each hazard type, through the generation of annual risk curves.
Specific risk results from multiplying the annual probability factor, vulnerability and cost or
indirectly multiplying annual probability with expected damage.
Specific risk maps were generated for each type of hazard and each return period by
multiplication of the potential damage maps and the annual exceedance probability. First
damage maps were generated by multiplication between vulnerability maps and cost
maps. For flood risk, damage maps were generated for three return periods (25, 50 and 75
years) from the various vulnerability maps multiplied by the cost map of the contents only,
because it was assumed that the flooding will normally have little influence on the building
itself. For the maximum flood damage map the vulnerability was considered to be 1 (total
destruction). For the lateral erosion damage map also a vulnerability of 1 was assumed,
since both buildings and their contents would be lost due to collapse in the event of
undercutting. Specific risk maps for seismic hazard were made for the four return periods
mentioned earlier (25, 50, 100 and 200 years), each using its own vulnerability map.
Estimation of specific risk for landslides was one of the most difficult tasks, since both the
probability, magnitude of the landslide, and therefore the vulnerability are very difficult to
predict. Here an expert judgment was made and three vulnerability classes were used: 0,
0.5 and 1.
The resulting specific risk maps gave information on the total amount of damage expected
annually due to a certain type of hazard with a certain return period. This damage was
aggregated for the entire city and plotted in a graph of probability versus potential damage,
61
though which a curve was fitted (e.g. Figure 2.22). The area below the graph represents
the total damage for the specific type of hazard. Out of these a total risk curve was derived
for the combination of the various hazard types, which represents the annual expected
losses to buildings and contents of buildings for the various types of natural hazards in the
city of Turrialba. The estimation of annual losses for each hazard type and each return
period represents a very important “standardization process” which allows hazards to be
put into perspective and prioritized accordingly. The data generated can also be used to
display a total risk map.
Figure 2.22 Specific risk curve for flooding (left) and seismic (right). X-axis is annual exceedance probability; Yaxis is estimated damage in Costa Rican currency
In summary, as added value we can say that in this study the same approach is used for
all the types of natural hazards considered. This makes the calculated risks for different
hazards comparable. The focus of the paper is to develop a GIS-based methodology for
risk assessment in urban areas. The GIS platform facilitates the communication of the
spatial distribution of risks and evaluation of the effectiveness of risk mitigation measures.
On the other hand, possible limitations of the methodology may be that, as with most of
those analysed in this review, it does not address the modelling of the interactions among
multiple hazards and elements at risk, or modelling of cascading events. Regarding the
relationship between magnitude and return period of the different events, as well as the
vulnerability assessment of the elements at risk, the analysis relies heavily on historical
information and expert judgment. The risk values estimated for individual buildings using
the approach outlined in the paper are at best indicative. But they do indicate the relative
importance of each type of hazard, and the degree of impact it is likely to have, for the
urban area as a whole.
2.3.9 The Central American Probabilistic Risk Assessment (CAPRA) approach
(http://www.ecapra.org);
CAPRA consists of a GIS-based platform for risk analysis, where probabilistic techniques
are applied to the analysis of earthquakes, tsunamis, hurricanes, floods, landslides and
volcanoes. Hazard information is combined with exposure and vulnerability data, allowing
the user to determine risk simultaneously on an inter-related multi-hazard basis,
distinguishing the platform from previous single hazard analyses.
62
Central to CAPRA’s innovation is the integration of Web 2.0 technologies. CAPRA is
fundamentally designed to be modular, extensible and open. This allows for mass
collaboration, and enables an ever-evolving and sustainable “living instrument”. Built upon
the platform, CAPRA applications consist of a risk map tool, a cost-benefit analysis tools
for risk prevention or mitigation, and programs that assist in the design of risk financing
strategies.
Probabilistic Risk Model Components:
In CAPRA, risk estimation is focused on probabilistic models that allow us to use the little
information available to predict possible catastrophic scenarios for which the uncertainty
involved in the analysis is considered high. Consequently, risk evaluation must follow a
prospective approach, anticipating scientifically probable events. Considering the large
uncertainties associated with the estimation of the magnitude and frequency of natural
disasters, the risk model is based on probabilistic functions that incorporate uncertainty in
the estimation of risk.
Hazard Module:
Hazard is understood as the definition of the frequency and severity of occurrence of a
determined danger at a specific location. Hazard analysis is based on the historical
frequency of events and the magnitude of each one of them. Once the hazard parameters
have been defined, it is necessary to generate a group of stochastic events which can
define the frequency and magnitude of thousands of events, representing the main hazard
parameters of the region. The analysis shows intensity parameter values defined for each
one of the studied hazards and for each one of the stochastic events that are being
considered, through modelling of each phenomenon. The distribution of the intensities
associated to adverse natural phenomena is an essential resource to evaluate risk.
Handling this type of information using raster layers in a GIS system, allows the
automation of risk calculation processes, as well as a simple and efficient way of sharing
the results.
Probabilistic models for the natural phenomena that are considered for the study are
presented. These include earthquake, hurricane, intense rainfall and volcanic action. In
this project, a multi-risk perspective is used, in the sense that it is possible to consider
different risks in an approach based on triggering phenomena; in this way, it is expected
that the analyst can capture the whole range of possible damage. In particular, the
triggering-based focus of the methodology gives the flexibility to determine (e.g. see Figure
2.23):
-
-
Total annual losses associated with a specific hazard (including losses associated
with all the considered secondary events). For example: total losses from hurricane
= sum of losses due to wind, tides, floods, and landslides;
Total losses associated with a specific category of hazard and related to different
triggering events, as for example: Total losses from landslides = sum of losses due
to landslides triggered by rain, earthquakes, hurricane-generated intense rains;
63
Figure 2.23 Modelling method of hazards based on triggering events
The result of the hazard assessment is a database for each of the analysed hazards,
which contain a set of exhaustive and mutually exclusive stochastic events, characteristic
of the ‘global’ hazard, that correspond to all possible hazard scenarios that can be
presented in the studied area. Each analysis provides a geographic distribution of the
affected areas using intensity parameters as outlined in Table 2.10.
Hazard
Earthquake
Effect
Mass movement
Earthquake
Tsunami
Hurricane
Hurricane-generated winds
Hurricane
Tide
Hurricane
Hurricane-generated rain
Non-hurricane rain
Floods
Landslides
Volcanic eruptions
Volcanic eruptions
Volcanic eruptions
Ash fall
Lava flows
Pyroclastic flows
Intensity parameter
Maximum
acceleration,
velocity, and displacement
(and spectral for different
structural periods)
Depth and area of flooded
area
Maximum
wind
speed
distribution (3 sec. gusts)
Depth and area of flooded
area
Depth
and
area
of
precipitation
Depth
and
area
of
precipitation
Depth and area of flooded
area
Distribution of security factor
or
landslide-susceptibility
indicator
Ash thickness distribution
Distribution of affected area
Distribution of affected area
Table 2.10 Intensity parameters considered of the different hazards
64
Exposure Module:
The exposure values of assets at risk are either estimated based on secondary sources,
such as existing databases, or derived using simplified procedures that use social and
macroeconomic general information, such as population density, construction statistics or
more specific particular information. These approximations are used when the specific
(asset by asset) information is not available. Based on the available information, an
exposure database is created, constructed in a geo-referenced manner, where the specific
required information is completed. Additional parameters, highly detailed, can be included
in order to improve the general reliability of the results. Special routines allow the
visualization of the information contained in the database and the calculation of general
interpretation indices.
In order to estimate the social impact of a certain event, the building occupation is
estimated. The maximum occupation and occupation percentages at different times of the
day are defined in order to perform analyses of scenarios with different times of
occurrence. When specific occupation information is not available, the approximate
occupation density of a certain construction type is used to complete said information.
For this study, the estimated exposure is calculated based on information from different
sources, such as current databases, asset inventory, land/property value information maps
and plans of use and territorial order. Additional information is generated from the analysis
of satellite image classification; these images are obtained as support for the study.
Finally, this information can be complemented with visits and local registries, which can
supply information for the analysis.
Vulnerability Module:
This module allows the user to assign vulnerability functions to each one of the assets
stated in the exposure inventory. Using the assigned vulnerability functions, it is possible
to quantify the damage or affectation produced in each of the assets before the action of a
certain event, characterized by one of the defined parameters of intensity.
The assignment of a vulnerability function to each asset is performed according to the
main characteristics of the component, in terms of its general behaviour regarding the
action of the considered event, as well as that of its main physical and mechanical
properties, which constitute the variable that defines its behaviour. The vulnerability
functions are calculated considering (1) the materials of the main structure and of
masonry; (2) general geometry; (3) structural type; (4) types of ends and junctions; (5)
date of building and normative used; (6) state evaluation and possible previous damages;
(7) defects and identifiable weakness ; (8) previous reinforcement works; and (9) inelastic
behaviour expected.
The estimation of damage or affectation is normally measured in terms of the Mean
Damage Ratio (RMD for its acronym in Spanish). The RMD is defined as the relationship
between the expected cost of repairing the affected element and the replacement cost of
the same element. The vulnerability function or curve is defined as the relationship
between the RMD and the intensity parameter that was selected for that particular event.
Each component of a given system will have a different vulnerability function assigned for
each one of the event to which it will be exposed.
65
Damage and Loss Module:
In order to calculate the losses associated with a certain event, the RMD, obtained from
the vulnerability function, and multiplied by the replacement value of the component,
represents the economic loss value. This operation is repeated for each asset or element
accounted for in the inventory for each one of the analysed events. Afterwards, the losses
are added up, following a proper arithmetic method that is suitable for probability density
functions, as required.
For loss calculations associated with risk transference and insurance, parameters such as
deductibles, maximum limits, coinsurances and others are taken into account.
The parameters that are used to measure the risk, give necessary information for future
risk management to those who are in charge of making the decisions. The main
measurements of economic risk are:
-
-
Expected Annual Loss (PAE): the sum, for all the considered events, of the
product of expected losses and the probability of occurrence of the given event;
Pure Risk Premium (PPR): the PAE divided by the replacement value of the asset;
Excess Loss Curve (CEP): annual frequency with which a given economic loss will
be exceeded. This is considered here as the most important measure for risk
management, since it provides basic information for planning and resources
assignment. The CEP may be estimated from the most probable event in a year, or
in a uniform way for all the possible events as function of the return period;
Probable Maximum Loss (PML): represents the loss value for a given
exceedance.
On the other hand, additionally to the probabilistic parameters of risk quantification that
have been mentioned, it is also useful for disaster management, emergency management
and risk mitigation, to have damage and loss scenarios from a deterministic perspective or
of specific events, such as some historical events or events with an established return
period.
Uncertainty considerations:
for each considered hazard, uncertainty is assessed in terms of the expected variability of
each key parameter by their respective coefficients of variation. This uncertainty is
considered at different levels of the modelling processes, up to arriving at a ‘global’
uncertainty assessment that can be assigned to the final values of estimated losses.
-Risk Evaluation from a Holistic Point of View:
Holistic risk evaluation is based on amplifying the physical risk by an impact factor that
depends on the specific conditions of the socioeconomic context in which the disaster
happens, which make the initial scenario of physical loss (Cardona and Hurtado 2000)
worse. The Urban Seismic Risk Index (USRi) proposed at Carreño et al., 2007, is a
composite indicator that measures the risk associated to earthquake from an integral
perspective, allowing risk management to include multidisciplinary aspects of vulnerability
reduction and intervention (e.g. see Figure 2.24). This approach can be applied to risk
evaluation for any type of natural hazard, from the initial evaluation of physical damage
66
with associated potential to the phenomenon. The social conditions that aggravate the
physical effects of the phenomenon are taken into consideration, doing so, the risk is
established for each analysis unit, through impact factors that modify the physical risk
associated with the vulnerability of buildings and infrastructure, in order to consider the
socioeconomic conditions of each unit.
Figure 2.24 Theoretical framework and model for holistic approach of disaster risk (from Carreño et al., 2007)
Once there is an evaluation for a certain village or city, it is simple to identify the most
relevant aspects in the total risk index. Mitigation priorities can be established in order to
modify those conditions (sub-indicators) that have a larger influence over the total risk.
This technique allows us to compare risk under different particular socioeconomic
conditions. It must be focused towards risk control instead of obtaining a precise
evaluation of it. The final objective is to supply the users with information that allows them
to take better decisions.
2.3.10 The ‘Regional RiskScape’ project in New Zealand: “Quantitative multi-risk
analysis for Natural hazards: a framework for multi-risk modelling.”
(Schmidt et al., 2011)
This paper introduces a generic framework for multi-risk modelling developed in the project
‘Regional RiskScape’ by the Research Organizations GNS Science and the National
Institute of Water and Atmospheric Research Ltd. (NIWA) in New Zealand. The goal was
to develop a generic technology for modelling risks from different natural hazards and for
various elements at risk. The technical framework is not dependent on the specific nature
of the individual hazard nor the vulnerability and the type of the individual assets.
The paper describes a generic framework and prototype software (called RiskScape) for
assessing the risk (loss) due to generic natural hazard(s). The fundamental components of
a risk model, namely hazard, values of assets (elements at risk) and [physical] vulnerability
models (fragility functions) for the system under consideration are specified in a
standardized manner and losses due to various natural hazard scenarios are evaluated.
67
Thus, the developed prototype system should be able to accommodate any hazard, asset
or fragility model, which is provided to the system according to that standard. The software
prototype was tested by modelling earthquake, volcanic ash fall, flood, wind, and tsunami
risks for several urban centres and small communities in New Zealand; in the paper,
example calculations are provided for earthquakes, storms and tsunami hazards at a
regional-scale (Hawke’s Bay in New Zealand), but the software could be applied at any
scale.
As of now the model only considers specific hazard event scenarios in a deterministic
manner. Neither possible interactions among the hazards nor cascading events are
considered. The authors claim that a probabilistic version of the software is under
development.
RiskScape functions for calculations of risk:
The functionalities of Riskscape are implemented as an application programmer interface
(API) based on the above RiskScape specifications and include the following:
1. Calculation of affected (exposed) assets. This involves either overlaying the spatial
representation of a hazard magnitude, the hazard exposures (e.g. flood depth,
ground shaking magnitude.) as derived from the hazard module onto the spatial
representation of asset module (e.g. building locations) as stored in the asset
module or the direct calculation of the same quantities based on a module-specific
algorithm. The calculation results in hazard magnitudes at asset locations (called
‘asset exposures’) hence affected assets are identified, and their exposure
determined.
2. Calculation of relative loss (damage ratio). This involves applying a user-selected
fragility function to a set of assets and asset exposures. The RiskScape system
extracts relevant asset attributes and asset exposures as required by the fragility
function, applies the fragility function and returns a damage ratio for each asset.
3. Calculation of absolute loss by including a valuation model for assets. If a valuation
is available for the considered asset (e.g. reconstruction costs, clean-up costs), the
valuation can be multiplied by damage ratio to return absolute damage or monetary
losses, respectively.
4. Calculation of time-averaged loss (=risk). If a hazard probability is available for the
used hazard module (and event), the absolute losses can be converted into losses
per time period—typically a year (e.g. as $/year). This enables the user to compare
potential losses for different event magnitudes.
5. Calculation of space-averaged loss can be of interest to planners to identify general
areas of higher potential impacts. This is done by applying an aggregation units
module to the calculated losses and returns losses per area (e.g. as $/km2 or as
$/km2/year).
RiskScape methodology was used to simulate the potential impacts of an extreme
earthquake event, a simulated design storm event and a simulated local river flood
scenario on the Hawke’s Bay building infrastructure. All the calculated damage ratios and
losses were aggregated onto a regular 1 x 1 km grid. This then allows the risk profiles of
the different hazards to be compared with each other. The presented results are indicative
only and are based on preliminary fragility models and asset estimates from available data
sources. The underlying asset and fragility databases do not allow an exact quantitative
68
risk calculation for the region. Moreover, some of the hazard models are preliminary in
nature. These results are presented only to illustrate the capabilities of the software
system and not as a quantitative risk study for the Hawke’s Bay area (e.g. see Figure
2.25).
Figure 2.25 Example output for a risk assessment for the Hawke’s Bay area using the RiskScape software. (a)
1931 historical earthquake event applied to calculate building losses in terms of reconstruction costs (scenaro);
(b) 1,000-year design storm event applied to calculate building damages, here expressed as damage ratio (%
destroyed); (c) Flood inundation from a Tutaekuri Meeanee river breach scenario applied to calculate building
damages, here expressed as damage ratio (% destroyed).
Comparison of risks for the different hazard scenarios can assist in cost–benefit analyses
of prevention measures and help targeting mitigation and emergency planning. All the
modelled scenarios are considered to be low-frequency extreme events. It has to be noted
that the comparison presented here is based on different event scenarios, and a more
robust multi-risk analysis needs to include a clear quantification of event probability, which
in this paper was not done, as information on event probability was available only for one
of the scenarios. The author argues that this is planned in future work.
Analysing the statistics of losses and the overall frequency distribution of calculated
individual losses (Figure 2.26) for the different hazard types delivers diagnostics of
individual hazard signatures, which -the authors argue- can be helpful for
planning/emergency purposes. They normalized loss as loss densities ($/km2) and
displayed them as cumulative statistics relative to the total affected asset base to make the
different scenarios comparable. The flood scenario mostly creates lower damages to
buildings in a confined area; the earthquake scenario, centred over Napier, affects many
high-value buildings, whereas the storm simulation creates low and high damages over a
much wider area; both the earthquake and the storm scenario create a similar large total
damage of about 3.5 billon NZ$ to the building stock.
69
Figure 2.26 Frequency distributions (% of affected area exceeding certain loss) of total building damage (as
reconstruction costs per km2) for the earthquake, flood and storm scenarios for Hawke’s Bay building assets.
The earthquake scenario has more higher damages
(~70% of damages>$10,000/km2; ~5% of
damages>$1,000,000/km2); the flood scenario has more lower damages (for ~50% of the area, damages are
negligible; for the rest of the area, the damages vary exponentially between $100,000/km2 and $100 Mio/km2);
the storm scenario has similar frequency characteristics compared to the earthquake scenario with higher
frequency of damages between $10,000/km2 and $1Mio/km2 (~80% of damages>$10,000/km2, almost 100% of
damages < $10 Mio/km2)
As added value, we can say that the same modelling approaches and methodology are
used for all types of natural hazards. This makes the calculated risks for different types of
hazard comparable. However, in its present form, the system is not capable of modelling
the interactions among multiple hazards and elements at risk, nor is it capable of modelling
cascading events.
2.3.11 Multi-risk initiatives in volcanic areas
Three initiatives of a kind of multi-risk assessment but considering just volcanic areas (a
single or multiple volcanic sources generating different kinds of volcanic-related hazards)
have been found: the “Explosive Eruption Risk and Decision Support for EU Populations
Threatened by Volcanoes” (EXPLORIS) project (e.g., Hincks et al., 2006); a volcanic risk
ranking initiative for the Auckland region, New Zealand (e.g. Magill and Blong, 2005a,b)
and the GRINP project at Mount Cameroon volcano (e.g. Thierry et al., 2008). Given that
the specific case of volcanic risk assessment (considered as a ‘single’ risk case) is
discussed with more detail in the MATRIX deliverable D2.1, here we give just a general
description of those documents and invite the reader to see a more detailed description of
some of those initiatives in the D2.1 of MATRIX.
In Hincks et al., 2006 (EXPLORIS), the probabilistic risk assessment methodology
proposed is based on an event-tree structure. It is stated that to enable a complete risk
assessment to be performed, multiple scenarios should be defined to represent the range
of possible eruptive behaviour of the volcano. A number of sub-events have therefore
been defined for each scenario, with appropriate distributions for eruption parameters and
event occurrence obtained using a combination of expert judgment and evidence from the
volcanic record.
70
Multiple hazard maps are required to capture the variation in extent and severity of impact
under each scenario/event. Where possible ensemble style hazard modelling can be
employed, using alternative numerical codes used to generate input hazard data. Here,
two methods are used to assign probabilities: (1) expert elicitation, and (2) statistical
distribution assignment.
Building damage is divided in to six categories, ranked in order of severity of risk to
building occupants:
Level 6: Structural failure due to pyroclastic flow;
Level 5: Structural failure due to earthquake;
Level 4: Roof collapse due to tephra loading;
Level 3: Failure of 3 or more windows or openings;
Level 2: Failure of 2 windows or openings;
Level 1: Failure of 1 window or opening;
Level 0: Building envelope intact (infiltration hazard only).
Vulnerability functions have been defined for all building classes to estimate probability of
damage under hazard conditions. Human vulnerability functions describe the probability of
death or serious injury to occupants. All vulnerability function parameters are assumed
uncertain, and allowed to vary with specified distribution functions.
The Event Tree formulation developed in EXPLORIS has been used to calculate the
probability-of-exceedance risk maps showing, for example, the annualised risk to building
stock. In EXPLORIS, risk it is proposed to be quantified in terms of causality numbers,
building loss, and individual risk.
In Magill and Blong (2005a,b), a method is presented for ranking volcanic hazards and
events according to their potential risk, and it is applied to the Auckland region. It is stated
that due to the long return periods of volcanic events and changes in eruption styles with
time it is often difficult to determine precise values for these parameters; then it is
suggested that a method for assigning values is to define categories for each parameter,
within which hazards can be placed, and then assign a value for each category. Due to the
large variability in volcanic hazard characteristics, they chose to define categories based
on order of magnitude values. This is analogous to a number of magnitude and intensity
scales used for classifying natural hazards (like the Richter magnitude scale for
earthquakes). In the case of volcanic eruptions, the Volcanic Explosivity Index (VEI)
contains nine categories with eruption volume increasing logarithmically from less than
104 m3 to greater than 1012 m3 (Newhall and Self, 1982). Similarly, resulting damage
ranges from small clean-up costs to total destruction and an order of magnitude scale also
seems appropriate.
The effect of a hazard is quantified based on the ‘susceptibility’ of the item affected and
the properties of the hazard; proportions of total loss for buildings and human life, within
the extent, were determined and effect values assigned. For simplicity, they assumed the
population to be evenly distributed throughout the affected area. For building damage,
‘order of magnitude’ categories for effect were based on the proportion of total building
loss. So that values could be more easily assigned to categories, a minimum proportion of
total loss was given to each category based on an average total replacement value of
NZ$200,000. Each hazard was assigned to the categories in a relative manner, based on
some observations that can be found in Magill and Blong (2005b).
71
Risk was calculated for building damage and loss of human life for every hazard caused
by each volcanic event, using the relationship (for details see MATRIX D2.1 and Magill
and Blong (2005a, 2005b):
Finally, in the GRINP project for Mount Cameroon (Thierry et al., 2008), The approach
follows the following steps:
- Characterization of the physical environment;
- Hazard assessment and zoning: hazard is estimated through the frequency (F) or
probability of occurrence of the different geological phenomena, the intensity (I) or
destructive power of the said phenomena and the extension (E) expressing what
percent of the threatened surface is possibly affected; Hazard threat index are
associated with global recommendation (e.g. « permanent human settlement is
inadvisable unless major precautions are taken »);
- Characterization of the exposed elements: inventory and localization of the exposed
elements (population, buildings, vegetation, strategic infrastructures, etc.);
- Vulnerability analysis of the various elements at risk: for each element at risk and
each type of phenomenon, a fragility curve is defined according to the methodology
exposed in Figure 2;
- Risk analysis : obtained by crossing information on hazard and vulnerability of
exposed elements
Figure 2.27 Diagram of the vulnerability assessment principle used in Thierry et al., 2006
As hazards, six volcanic phenomena linked to the eruptions (lava flows, aerial fallout, gas,
lahars, medium- and large-amplitude ground movements, tidal waves), two slope instability
phenomena (landslides and block falls), and one tectonic phenomenon (earthquakes).
As added value of this project, it can be mentioned that a practical and structured
approach to combine various hazards affecting the same area was proposed. An
approach to assess and analyse risk is presented, which has implied bringing together and
structure in a seamless way much information on exposed elements and their vulnerability
to each hazard. Finally, this study served as basis for the Cameroonian authorities for
implementing a hazard prevention strategy and for improving preparedness to crisis.
72
3. Summary of the main results: general maps, risk indices and risk
curves
In section 2 we summarized the methodological approaches proposed by the most
important European and international multi-risk projects, as well as research papers and
practical applications found in literature. In this section we will summarize the most
important results and achievements as outlined by this state-of-the-art analysis, as well as
the main constraints and gaps found.
3.1
General considerations
Starting from its general definition (e.g. see. MATRIX deliverable D3.2), multi-risk
assessment may be understood as the process “to determine the whole risk from several
hazards, taking into account possible hazards and vulnerability interactions”. Then, a full
multi-risk approach entails both a multi-hazard and multi-vulnerability perspective. The
kind of hazardous events considered may include (1) events occurring at the same time or
shortly following each other (because they are dependent on one another or because they
are caused by the same triggering event or hazard) –which is mainly the case of
“cascading events”—, or (2) hazardous events threatening the same elements at risk
(vulnerable/exposed elements) without chronological coincidence (e.g. European
Commission, 2010).
From the bibliographic review performed, it emerges that most –if not all- of the initiatives
on multi-risk assessment have developed methodological approaches that consider the
multi-risk problem in a partial way, since their analyses basically concentrate on the
second part of the general definition, i.e. the risk assessment for different hazards
threatening the same exposed elements. Within this framework, the main emphasis has
been towards the definition of procedures for the homogenization of spatial and temporal
resolution for the assessment of different hazards. For vulnerability instead, being a wider
concept, there is a stronger divergence on its definition and assessment methods;
considering physical vulnerability issues, a more or less generalized agreement on the use
of vulnerability functions (fragility curves) has been reached, which facilitate the application
of such a kind of multi-risk analysis, however, for other kinds of vulnerability assessment
(e.g. social, environmental, etc.) it is less clear how to integrate them within a multi-risk
framework.
In this framework the final multi-risk index is generally estimated as a simple aggregation
of the single indices estimated for different hazards. Other approaches consider a single
hazard at a time and multiple exposed elements (e.g. buildings, people, etc.) for the
vulnerability, which are combined and weighted according to expert opinion and subjective
assignment of weights.
The choice of the methodology strongly depends on both the scale of the study and the
availability of information (for hazard and vulnerability assessment). Noteworthy, many of
the approaches found define theoretical frameworks for the multi-risk assessment that,
when applied to real cases, are generally simplified This is due to the difficulty in obtaining
the detailed information needed. It is also interesting to point out that many of the reports
discuss the importance of the interaction among hazards and cascading of events for a
fully multi-hazard perspective, however, few efforts have been done in order to define a
rigorous methodology.
73
3.2
Scale constraints
The first point that we can outline is about the scale constraints. Looking at the most
important applications, it is evident that the methodological approach used is strongly
determined by the scale of the study; for instance, if we consider the ‘large-scale’
methodologies (see Sec. 2.1), the multi-risk analysis is generally performed by the use of
risk indices representing expected annual mortality and economic losses; a “total” risk
index is estimated as a simple aggregation of single risks, and hazard or vulnerability
interactions or cascade effects are not considered. This kind of result represents a
synoptical methodology principally addressed to global policies with very low reliability at
the local scale; the objective being to identify hotspots where natural hazard impacts may
be largest.
As we go down in the scale of the analysis (see Sec. 2.2 and 2.3), multi-risk assessment is
generally based on more detailed analysis. For this kind of procedures, risk from different
hazards is quantified either using a common metrics (in general expected mortality or
economic losses in a given timeframe –normally 1 year-), or based on normalized indices
resulting from the grouping of hazard intensities and vulnerability degrees in generic
classes (low to high). The results are generally expressed using risk curves or risk indices
that, as result of homogenized analysis, may be ranked and allow direct risk comparison
for different typologies of natural and man-made adverse events.
3.3
Presentation of results: general risk maps, risk curves and risk indices.
Among all the methodologies proposed in the reviewed project reports and research
papers, different approaches to represent the final results have been proposed. As
mentioned before, the kind of analysis strongly depends on the scale of the particular
problem, hence the different possible output results.
Figure 3.1 summarizes some of the most important typologies of results proposed in the
reviewed multi-risk methodologies. From the global methodologies, the output of the multirisk analysis is generally presented as global maps (discretized in grids of different spatial
resolution) of expected annual losses in terms of the metric chosen for the analysis (i.e.
mortality or economic), as reported for example in Figure 3.1a. Risk indices, in terms of a
given quantitative loss metric or normalized and/or relative index values, are generally
used in (medium) regional scale studies. An example of this kind of results is presented in
Figure 3.1b from the ESPON project, in which an aggregated risk map (generated from 15
aggregated natural and anthropogenic hazards) is presented using an ad-hoc index. In this
case, the final map is an integrated risk map that is a combination of the vulnerability map
and the aggregated hazard map (based on a weighting procedure using the Delphi
method). This kind of risk map is accompanied by a legend that displays the hazard values
on the y-axis and the integrated vulnerability on the x axis, and where different shades of
the same colour allow distinguishing between a higher intensity of a hazard or a higher
degree of vulnerability, respectively.
An example of a risk index based on a given quantitative loss metric is presented in
Figure 3.1c. In this case, different scenarios-damage indicators are used to determine
different scenario-based risk indices, which after weighting are summed up to determine a
final aggregated risk index.
74
The use of risk curves for the results of risk quantification is more common in local scale
analysis, i.e., the risk curves are the result of the combination of a generic hazard curve
(that relates a given hazard intensity measure in the x axis with the return period,
probability of occurrence or probability to threshold overcome in the y axis), with one or
more fragility curves (that relate the same hazard intensity measure with the probability of
damage, allowing the translation of vulnerability to loss). Two possible approaches have
been found in literature to use risk curves in a multi-risk analysis. Figure 3.1d is a sketch
proposed by ARMONIA (2007), in which different kinds of vulnerability functions are
plotted having a different x axis for any hazard typology and a common y axis (e.g., an
average damage metric). Therefore, it would be possible to represent all the risk with the
same factor; in this way a harmonized conjunction point between hazards and risk can be
found.
On the other hand, comparative plots of different risk curves can be also constructed as
shown in Figure 3.1e; in this kind of approach, the risk curve relates the (generally annual)
probability of exceedance (of a given loss) on the y axis, with the losses (in the considered
metric) on the x axis. The loss estimation through a common loss unit (in this case chosen
to be the direct economic losses) makes the risks associated with different natural hazards
quantitatively comparable. These kinds of risk curves may be a mid-way step before
arriving at a final risk index useful for risk ranking. For instance, when integrating the risk
curves it is possible to obtain a single risk value (with the same units of measure) that can
be used for ranking the risks, as shown in Figure 3.1f, which represents the risk ranking for
Casalnuovo municipality (Marzocchi et al., 2009).
75
(a)
(b)
(c)
(d)
(e)
(f)
76
Figure 3.1 Summary of the most important typologies of results of multi-risk methodologies found in literature:
(a) example of global distribution of flood risk (expected mortality, from Dilley et al., 2005) ), for details see
Figure 2.2 in this document; (b) example of an aggregated risk map from the ESPON methodology (from
Schmith-Thomé, P. Ed., 2005); for details see Figure 2.4 in this document; (c) combination of scenarios (i) and
damage indicators (j) to determine the risk indices Rj,i (from Carpignano et al., 2009), for details see Figure 2.18
and Figure 2.19; (d) diagram of a proposed ARMONIA function between hazard types intensity (x-axis) vs.
average damage of exposed elements (y-axis) according to scale of analysis (from ARMONIA, 2007, for details
see Figure 2.6 of this document); (e) comparison of risk curves for the hazards due to windstorms, floods and
earthquakes (from Grünthal et al., 2006), for details see Figure 2.14 in this document; (f) risk ranking (from
Marzocchi et al., 2009 [Note: the sketch of the risk curves is from ARMONIA, 2007]).
3.4
Final remarks, limitations and gaps
The analysis performed of the state-of-the-art on multi-risk assessment from the different
European and international initiatives described in this document has highlighted the
following main features and gaps to be considered for the development of MATRIX
products:
-
Different methodologies, ranging from simplified approaches to innovative and
advanced methods, were identified in this state-of-the-art analysis. Nevertheless,
practically all the reported studies present important problems when transformed
into practical applications. For instance, any methodological approach for multi-risk
assessment is strongly constrained by both data availability (for hazard and
vulnerability assessment) and the scale of the problem;
-
Multi-risk approaches may imply multiple hazards affecting the same exposed
elements, and/or one or more hazard affecting different categories of exposed
elements. In the first case, quantitative risk assessment is generally more viable
since a common metric for loss assessment is easier to be defined (i.e. risk
harmonization based on the harmonization of effects). In the second case,
considering different categories of exposed elements (e.g. buildings, population,
green areas, environmental, etc.) imply difficulties on both the definition of a
common metric for loss assessment, and how to weight the different categories of
exposed elements. This kind of analysis involves strong subjective decisions that
are not always easy to justify, and the risk quantification is generally performed
using normalized indices that may allow, for example, individuate hotspots of high
risk to be identified. However, sometimes its utility for risk management and
decision-making may be questionable.
-
The most basic requisite for a quantitative multi-risk assessment is the definition of
a target area, common time frame, a quantitative assessment of hazards (generally
in probability terms), a coherent vulnerability assessment (i.e., linked to the intensity
measure parameterizations adopted for the hazard assessment), and a defined
metric to quantify losses. However, the choice of a specific kind of loss metric may
present different problems and limitations. In fact, the effect of different hazards
may have different temporal characteristics (e.g. the recovery of construction is not
the same of that of agricultural land or trees). Also different return periods, for
77
different hazards, may pose difficulties to integrate the cost over a given period of
time.
-
A strong limitation found up to now is that none of the analysed studies produce a
rigorous methodology for multi-hazard assessment. Most of the multi-risk
methodologies consider the effect of different hazards as independent, neglecting
the possibility of hazard interaction or cascade effects. Worthy of note, many of the
reviewed documents write comments about the importance that cascades of events
or hazard interactions may have for the risk, but very few try to quantify some basic
scenarios.
-
Linked to the previous point, multi-risk assessment requires also a careful
evaluation of the interaction between vulnerabilities to different hazards. For
example, the seismic vulnerability of an edifice changes significantly if the roof is
loaded by volcanic ash. Only very little effort has been devoted to tackle this issue.
-
One of the main gaps found for the practical application of the more important
quantitative multi-risk methodologies found in literature is the lack of fragility curves
derived by intensity (of the hazardous event) vs. typology of exposed elements.
This topic can be considered as one of the most significant matters to be addressed
for future developments of multi-risk analysis, especially in high-resolution analysis
(at local scale).
-
Another gap found is in the treatment of uncertainties. None of the methodologies
consider uncertainty quantification at any step of the process (except for some
specific hazard assessment approaches), nor propagate (epistemic) uncertainties
up to the final risk values.
78
4. References
ARMONIA: Applied multi-risk mapping of natural hazards for impact assessment, 2007.
EC project, contract 511208. (Consulted deliverables: 3.1, 3.1.1, and 6.1)
Bovolo C.I., S.J. Abele, J.C. Bathurst, D. Caballero, M. Ciglan, G. Eftichidi, and B. Simo,
2009. A distributed framework for multi-risk assessment of natural hazards used to model
the effects of forest fire on hydrology and sediment yield. Computers & Geosciences, 35:
924-945. DOI: 10.1016/j.cageo.2007.10.010
Carnec, C., H. Modaressi, J. Douglas, D. Raucoules, and E. Simonetto, 2005. Contribution
of space imagery to vulnerability assessment of elements exposed to geological risks.
TS14 Disaster II/General, 31st International Symposium on Remote Sensing of
Environment: Global Monitoring for Sustainability and Security, St Petersburg, Russian
Federation, 20-24 June.
Cardona, O.D., and J.E. Hurtado, 2000. Modelación numérica para la estimación holística
del riesgo sísmico urbano, considerando variables técnicas, sociales y económicas. En
Métodos numéricos en ciencias sociales, editado por E. Oñate, F. García-Sicilia y L.
Ramallo. pg. 452-466. International Centre for Numerical Methods in Engineering
(CIMNE), Barcelona. ISBN: 84-89925-71-2.
Carpignano, A., E. Golia, C. Di Mauro, S. Bouchon, and J-P. Nordvik, 2009. A
methodological approach for the definition of multi-risk maps at regional level: first
application. Journal of Risk Research, 12 (3-4): 513-534
Carreño, M.L., O.D. Cardona, and A.H. Barbat, 2007. Urban seismic risk evaluation: A
holistic approach, Nat. Hazards, 40:137-172 DOI: 10.1007/s11069-006-0008-8
Climent A.M. 1997. - Evaluacion de la amenaza sismica: terremotos, intensidades,
aceleraciones, espectros y fenomenos secundarios. 1er. Taller C.Am. para la reduccion
de los efectos de las amenazas sismica en las subestaciones electricas. I.C.E.
Dilley, M., R.S. Chen, U. Deichmann, A.L. Lerner-Lam, M. Arnold, (with: J. Agwe, P. Buys,
O. Kjekstad, B. Lyon and G. Yetman), 2005. Natural Disaster Hotspots: A global risk
analysis. The world Bank hazard management unit, Washington, D.C.
Del Monaco, G., C. Margottini, and S. Serafini. 1999. Multi-hazard risk assessment and
zoning: an integrated approach for incorporating natural disaster reduction into sustainable
development. TIGRA (The Integrated Geological Risk Assessment) Project (Env4-CT960262) Summary Report.
Douglas, J., 2005. RISK-NAT (Module 4): Methods and tools for risk evaluation. Progress
report. BRGM/RP-54041-FR. http://www.brgm.fr/publication/rapportpublic.jsp
Douglas, J., 2007. Physical vulnerability modelling for natural hazards risk assessment.
Natural Hazards and Earth System Sciences, 7(2), 283-288. http://www.nat-hazards-earthsyst-sci.net/7/283/2007/nhess-7-283-2007.html
European Commission DG XII, Environment and Climate Program, 2000. TEMRAP: The
European Multi-Hazard Risk Assessment Project, contract ENV4-CT97-0589.
European Commission, 2010. Commission staff working paper: “Risk assessment and
mapping guidelines for disaster management”, Brussels, December 2010
79
FEMA, 1997. Multi hazard identification and risk assessment: a cornerstone of the national
mitigation strategy, 1th Edition, United States.
http://www.fema.gov/library/viewRecord.do?id=2214
FEMA, 2001. Understanding your risks: identifying hazards and estimating losses, FEMA
REPORT 386-2, August 2001. http://www.fema.gov/library/
Granger, K., and T. Jones, 2001. A Multi-hazard risk assessment, in Granger, K. and M.
Haine (editors), Natural Hazards and the Risk they Pose to South-East Queensland.
Geoscience Australia, Canberra
Granger, K. and M. Haine (editors), 2001. Natural Hazards and the Risk they Pose to
South-East Queensland. Geoscience Australia, Canberra.
Grünthal, G., A.H. Thieken, J. Schwarz, K. Radtke, A. Smolka, and B. Merz, 2006.
Comparative risk assessment for the city of Cologne, Germany – storms, floods,
earthquakes, Nat. Hazards, 38(1–2), 21–44.
Hincks, T., G. Woo, and W. Aspinall, 2006. Probabilistic risk assessment methodology and
software for EXPLORIS volcanoes (Del. D6.2). EXPLORIS: Explosive Eruption Risk and
Decision Support for EU Populations Threatened by Volcanoes. EC project number EVR1CT-2002-40026.
Hollenstein K., 2005. Reconsidering the risk assessment concept: Standardizing the
impact description as a building block for vulnerability assessment. Natural Hazards and
Earth System Sciences, 5, 301–307. http://nat-hazards-earth-systsci.net/5/301/2005/nhess-5-301-2005.pdf
Kleist, L., A.H. Thieken, M. Müller, I. Seifert, D. Borst, and U. Werner, 2006. Estimation of
the regional stock of residential buildings as a basis for comparative risk assessment for
Germany. Nat. Hazards Earth Syst. Sci., 6, 541-552 http://www.nat-hazards-earth-systsci.net/6/541/2006/nhess-6-541-2006.pdf
Laporte M., C. Lindholm, H. Bungum, and Dahle A., 1994. Seismic Hazard for Costa Rica.
NORSAR, technical report No. 2-14, 73 pp.
Lari, S., P. Frattini, and G.B. Crosta, 2009. Integration of natural and technological risks in
Lombardy, Italy. Nat. Hazards Earth Syst. Sci., 9: 2085–2106.
Marzocchi, W., M.L. Mastellone, S. Di Ruocco, P. Novelli, E. Romeo, and P. Gasparini,
2009. Principles of multi-risk assessment: interactions amongst natural and man-induced
risks, Project Report (FP6 NARAS project), European Commission, Directorate-General
Research – Environment, contract No. 511264.
Marzocchi W, L. Sandri, P. Gasparini, C. Newhall, and E. Boschi, 2004. Quantifying
probabilities of volcanic events: the example of volcanic hazard at Mt. Vesuvius. J
Geophys Res 109:B11201. DOI 10.1029/2004JB003155
Marzocchi W., L. Sandri, and J. Selva, 2008. “BET_EF: a probabilistic tool for long- and
short-term eruption forecasting”, Bull. Volcanol., 70, 623-632, doi: 10.1007/s00445-0070157-y, 2008.
Magill, C., and R. Blong, 2005(I). Volcanic risk ranking for Auckland, New Zealand. I:
Methodology and Hazard investigation. Bull. Volcanol. 67: 331-339. DOI: 10.1007/s00445004-0374-6
80
Magill, C., and R. Blong, 2005(II). Volcanic risk ranking for Auckland, New Zealand. II:
Hazard consequences and risk calculation. Bull. Volcanol. 67: 331-339. DOI:
10.1007/s00445-004-0374-6
Merz, B. and A.H. Thieken, 2009. Flood risk curves and uncertainty bounds, Nat. Hazards,
51(3), 437–458. http://www.springerlink.com/content/v7vn684343684412/
Munich Re: http://www.munichre.com (section Publications): e.g. NATHAN worldmap of
natural hazards, 5th. Edition (2011) – last visited: June 2011
OCHA - United Nations Office for Coordination of Humanitarian Affairs, OCHA regional
office for Asia and the Pacific, 2009. Risk assessment and mitigation measures for natural
and conflict related hazards in Asia-Pacific. Norwegian Geotechnical Institute (NGI) report
20071600-1
Piedmont Region (2004) Linee guida per la redazione dei piani comunali di Protezione
Civile, Regione Piemonte.
Sauter, F. and H.C. Shah, 1978. Studies on earthquake insurance, Proceedings of the
Central American Conference on Earthquake Engineering, V2 San Salvador, El Salvador.
Schmidt, J., I. Matchman, S. Reese, A. King, R. Bell, R. Henderson, G. Smart, J. Cousins,
W. Smith, and D. Heron, 2011. Quantitative multi-risk analysis for natural hazards: a
framework for multi-risk modelling, Nat. Hazards, DOI 10.1007/s11069-011-9721-z
Smith, D.I. 1994. Flood damage estimation – A review of urban stage -damage curves and
loss functions. Water SA, 20(3), 231-238.
Schmidt-Tomé, P., (Editor), H. Kallio, J. Jarva, T. Tarvainen, S. Greiving, M. Fleischhauer,
L. Peltonen, S. Kumpulainen, A. Olfert, J. Schanze, L. Bärring, G. Persson, A.M. Relvão,
M.J. Batista, 2006. The Spatial Effects and Management of Natural and Technological
Hazards in Europe (ESPON) project 1.3.1. Geological Survey of Finland
(http://www.espon.lu – last visited: July 2011).
Trifunac M.D., and A.G. Brady, 1975. Correlations of peak acceleration, velocity and
displacement with earthquake magnitude and site condition: Intl. Jour. Earthquake Eng.
Struc. Dynamics. V 4. p 455-471.
United Nations Development Program (UNDP), 2004. A Global report - Reducing disaster
risk: a challenge for development. New York, Bureau for Crisis Prevention and Recovery.
Van Westen, C.J., L. Montoya, L., and L. Boerboom, 2002. Multi-hazard risk assessment
using GIS in urban areas: A case study for the city of Turrialba, Costa Rica. Proceedings
of the Regional Workshop on Best Practices in Disaster Mitigation, pp 53-72.
http://www.adpc.net/audmp/rllw/themes/th1-westen.pdf
81
82