A novel approach to flood risk assessment : the Exposure-Vulnerability matrices

The classical approach to flood defence, focused on reducing the probability of flooding through hard defences, has been gradually substituted by flood risk management approach, which accepts the idea of coping with floods, and aims at reducing both probability and the consequences of flooding. In this view, the concept of vulnerability becomes central, such as the (non-structural) measures for its increment. However, the evaluations for the effectiveness and methods of non-structural measure and the vulnerability are less studied, compared to the structural solutions. In this paper, we adopted the Longano catchment in Sicily, Italy, as the case study. The methodology developed in the work enabled a qualitative evaluation of the consequences of floods, based on a crisscross analysis of vulnerability curves and classes of exposure for assets at risk. A GIS-based tool was used to evaluate each element at risk inside an Exposure-Vulnerability matrix. The construction of an E-V matrix allowed a better understanding of the actual situation within a catchment and the effectiveness of non-structural measures for a site. Referring directly to vulnerability can also estimate the possible consequences of an event even in those catchments where the damage data are absent. The instrument proposed can be useful for authorities responsible for development and periodical review of adaptive flood risk management plans.


Introduction
The concept of risk implies a transition from the classical approach of defending a territory from flood hazard, through structural measures that modify the characteristics of the flood event, to the approach of reducing flood risk, through structural and non-structural measures that act on both flood hazard and its consequences.The EU Directive underlines the importance of prevention-oriented approaches, adopting early-warning systems, flood forecasting techniques, and land use regulation.
The use of prevention measures that do not interfere RQ IORRG ¶V IHDWXUHV UHTXLUHV WKH HODERUDWLRQ RI methodologies and strategies to verify their effectiveness.All over the world, public governmental bodies and academics published some studies on the effectiveness of non-structural measures [1±3], but the lack of data on it (or their coarseness) makes their reliability hard to know.The variable in risk equation [4] that describes the attitude of a territory in suffering impact of an hazardous event is vulnerability, GHILQHG DV ³WKH FKDUDFWHULVWLFV DQG circumstances of a community, system or asset that make LW VXVFHSWLEOH WR WKH GDPDJLQJ HIIHFWV RI D KD]DUG´ [5].7KH WHUP ³YXOQHUDELOLW\´ KDV DFWXDOO\ GLIIHUHQW interpretations, as existing epistemological traditions in various research areas with different objectives [6±8].Fuchs et al. [9] summarised these definitions of vulnerability with respect to natural hazards research.From a natural science perspective, studies on vulnerability focus on the susceptibility of physical systems in areas at risk to natural processes.Vulnerability is therefore considered as loss degree or percentage of damage that assets in areas at risk may suffer, which depends not only on hazard attributes, but also on the intrinsic characteristics of the affected element.
The definition of vulnerability is often confused with the one of exposure, which is defined as the number of assets being present in endangered areas distinguished per typologies [5].Studies related to flood vulnerability assessment are few because the uncertainty involved is difficult to quantify.In fact, although different damage assessment methods have been developed [10±14], the lack of high-quality essential data remains as the main obstacle to the derivation of uncertainties in ex-ante analysis.

Methodology
In this paper, we have developed a new methodology to assess flood risk for buildings based on their exposure classes and the relationship between flood depth and vulnerability.The goal is to describe flood consequences or flood risk in watersheds where vulnerability data do not exist or their quality makes them unreliable.
We considered the relation of vulnerability as the impact of natural hazards, generally measured in terms of damages or losses, and assessed the vulnerability as the expected loss degree of an element (or set element) at risk as a consequence of a hazardous event [15,16].Meanwhile, we further normalised the vulnerability to a value between 0 and 1, as the expected degree of loss varying from no damage to complete disruption.On the contrast, we regarded exposure as the pure identification of assets at risk and attributed the nominal value of each element based on the function of its strategic, economic and functional role.No monetary value was associated to buildings.
Our methodology includes four steps.At first, a hydraulic modelling is applied to derive the hydrodynamic characteristic of the flood event studied.The second step is determining the Exposure based on the building categories provided in the Flood Risk Plan for Sicily [17].The third step is deriving vulnerability curves for different buildings in Sicilian territory through a synthetic approach.Finally, the vulnerability assessment for different Exposure categories, referring to a flood event occurred in Barcellona Pozzo di Gotto (North-East Sicily, Italy): the results have been reported in an Exposure-Vulnerability matrix, allowing an immediate understanding of flood consequences.

Hydraulic modelling
To simulate flood propagation, a 2D model (Aronica et al., 1998)  where H(t,x,y) is the free surface elevation, u and v are the flow velocity components in x and y directions, respectively, h is the depth of flow, J x and J y are the friction terms in the x and y directions.
The friction terms are represented through the classical Manning-Strickler formulation as: The equations are solved by using a finite element technique with triangular elements to reproduce the complex topography of the built-up areas.More details on the model can be found in Aronica et al. [18].

'HILQLWLRQ RI ([SRVXUH ¶V FODVVHV
We adopt Exposure as a global estimation of EXLOGLQJV ¶ VRFLR-economic situation to flood risk, which depends on the property value, the social function, the indirect involvement in economic losses and the population density of the neighbourhood area.
Table 1 shows the building categories based on the above mentioned variables.The main category refers to the membership class considered in the Sicilian Flood Risk Plan.The classification is further refined for residential and public buildings according to their economic or strategic value, which are denoted as second index in the table.The third index describes the detailed type under sub-categories, which are classified depending on their economic or strategic value.This level of detail enables to perform a vulnerability analysis at building scale.

Definition of vulnerability curves for buildings
The basic idea of this study was the derivation of relative vulnerability functions for those sites where both damage data and on-site building inspections are lacking.
Final aim in the derivation of vulnerability curves was to describe possible damages occurring after fluvial floods in urbanized area and to make the curves as generic as possible.While referring to fluvial floods, often characterized by low velocities, another initial condition was to neglect structural damages to the buildings and to consider what happens to non-structural building components.
The first step in synthetic approach is to introduce the building typologies for which derive the curves: buildings are usually distinguished at first in function of their use, than in function of their structural features (e.g.materials, numbers of floors, extension, geometry, age, etc.).This LPSOLHV VWURQJ K\SRWKHVLV RQ WKH EXLOGLQJV ¶ VWUXFWXUH DQG the incorporation of each building presented in the study areas inside these standard pre-defined models.
We considered the damage of non-structural building elements and hypothesized the substitution cost of each element to derive its weight respect to the total substitution costs.To describe the proportional damage relative to each element, a questionnaire was distributed to a team of experts, in particular a team of civil engineers working in Sicily area.
The first step of the analysis consists of deciding ZKLFK EXLOGLQJV ¶ FODVVHV WR be included: this distinction is just referred WR WKH EXLOGLQJV ¶ W\SH EHFDXVH WKHLU IXQFWLRn has been already considered through their exposure.The same curves can be used for buildings with the same constructive features, even if they have different functions, such as residential or commercial.On the other side, different curves should be used for buildings with the same functions but with different constructive features.
We associated concrete buildings without basement to three finishes types: rich finishes for the building types such as villas and cottages; medium finishes for flats and single houses inside towns; and poor finishes for detached houses and single houses in villages.
After defining these conditions, we analysed the damage of finishes components that included floors, walls, doors and French windows, windows, wiring, water plant, gas plant and services.Their substitution prices were taken from the official price lists and depend on their quality and materials, which in turn were derived IURP WKH ILQLVKHV ¶ FODVV )RU H[DPSOH GRRUV LQ SRRU houses were hypothesized to be hollow wooden, while in rich ones were supposed to be in solid wood: these led to different substitution costs that weighed differently in respect to the total costs.These components could also suffer different damages for the same water depths due to the duration of flooding.
Once that all these conditions are defined, a team of experts was asked to describe, according to LQGLYLGXDO ¶V experience, how each component suffer damages in all the illustrated structures: the results were used to build a VHULHV RI ³SDUWLDO´ YXOQHUDELOLW\ FXUYHV RQH IRU HYHU\ building element in a particular combination of finishes class and event duration.
The sums of the partial curves that were related to the elements of a building type, each one multiplied for its weight, produced two total vulnerability curves for each building type: one for short and one for long duration hypothesis.
A separate effort needs to be done regarding the vulnerability curves for commercial activities: the majority of them are located in structures with the same materials and building characteristic of residential constructions: the same vulnerability curves can be so used, because in the general analysis their exposure class will play the role to distinguish them from each other.
While considering supermarkets and stores, the stock contents have higher weights in the damage estimation.For these typologies, a double distinction has been made: on one side, they have their own exposure class; on the other side, a vulnerability range varying linearly from 0 to 1, while the water depths vary from 0 to 60 cm has been considered.The reason for this last choice was due to the fact that it seemed plausible that when the water depth reach the height of 60 cm, the goods and the machineries (like fridges) contained in supermarkets and stores would be so damaged that a vulnerability value equal to 1 can be associated to them.

Vulnerability assessment
As previously introduced, the input data used for direct impact assessment are the flood inundation depths LQVLGH WKH EXLOGLQJV WKH EXLOGLQJV ¶ H[SRVXUH FODVVHV DQG the vulnerability curves.Flood inundation depths under various scenarios can be obtained using 2D hydraulic modelling.Exposure classes can be mapped at microscale (i.e., individual buildings) or at larger scales as land cover classes but, given the detail in exposure classification, the relationship between land cover class DQG EXLOGLQJV ¶ XVH VKRXOG EH GHVFULEHG To analyse the flood impact for individual buildings efficiently, we adopted the tool developed by Chen et al. [19]  The results for each flood condition is further summarised to highlight the severity in different exposure category and help decision making.Considering the uncertainties associated with the vulnerability within a catchment, a banded severity in an Exposure-Vulnerability matrix, as shown in Table 2, was used instead of simple one-to-one relationship curves.
The severity in each exposure type is banded up to five classes (low, moderate, medium, high, extreme), according to the function and importance of the type.For example, hospital would have less tolerance to hazard such that the severity was banded as high even the vulnerability was low.On the contrast, sparse houses could cope with more extreme conditions such that the severity was classified as high for the highest vulnerability class.
based on De Saint Venant equations has been used.
that can assess the vulnerability of each element for multiple flood conditions.