Seismic vulnerability assessment using field survey and remote sensing techniques

B. Murgante et al. (Eds.): ICCSA 2011, Part II, LNCS 6783, pp. 109–124, 2011. © Springer-Verlag Berlin Heidelberg 2011

Seismic Vulnerability Assessment Using Field Survey and Remote Sensing Techniques

Paolo Ricci1, Gerardo Mario Verderame1, Gaetano Manfredi1, Maurizio Pollino2, Flavio Borfecchia2, Luigi De Cecco2, Sandro Martini2, Carmine Pascale3,

Elisabetta Ristoratore3, and Valentina James3

1 Department of Structural Engineering (DIST) - University of Naples Federico II Via Claudio 21, 80125 Naples, Italy

2 ENEA - National Agency for New Technologies, Energy and Sustainable Economic Development - “Earth Observations and Analyses” Lab (UTMEA-TER)

Casaccia Research Centre - Via Anguillarese 301, 00123 Rome, Italy 3Consorzio T.R.E. Tecnologie per il Recupero Edilizio

Via Privata D. Giustino 3/A, 80125 Naples, Italy {paolo.ricci,verderam,gamanfre}@unina.it

{maurizio.pollino,flavio.borfecchia,luigi.dececco, sandro.martini}@enea.it

{carmine.pascale,elisa.ristoratore, valentina.james}@consorziotre.it

Abstract. In this paper, a seismic vulnerability assessment at large scale is de-scribed, within the SIMURAI project. A field survey was carried out in order to gather detailed information about geometric characteristics, structural typology and age of construction of each single building. An airborne Remote Sensing (RS) mission was also carried out over the municipality of Avellino, providing a detailed estimate of 3D geometric parameters of buildings through a quite fast and easy to apply methodology integrating active LIDAR technology, aeropho-togrammetry and GIS techniques. An analytical seismic vulnerability assess-ment procedure for Reinforced Concrete buildings is illustrated and applied to the building stock considering (i) field survey data (assumed as a reference) and (ii) LIDAR data combined with census data as alternative sources of informa-tion, according to a multilevel approach. A comparison between the obtained results highlights an acceptable scatter when data provided by RS techniques are used.

Keywords: Seismic vulnerability, Large scale, GIS, Airborne LIDAR, 3D Buildings.

1 Introduction

The SIMURAI Project (Strumenti Integrati per il MUlti Risk Assessment territoriale in ambienti urbani antropizzatI, Integrated tools for large scale multi-risk assessment in urban anthropic environment), co-funded by the Italian Ministry of Research (MIUR), aims at the definition of integrated tools for multi-risk assessment in highly

110 P. Ricci et al.

urbanized areas. Integrated assessment procedures for seismic risk are based on al-ready existing census data integrated with data acquired through innovative, fast and easy to apply techniques.

Different data acquisition techniques have been applied for the construction of GIS layers dedicated to buildings and streets description in the Decision Support System (DSS) developed for the project: existing cartography, on-site direct measures through specific input interfaces for Tablet PC, airborne LIDAR technology and Mobile Map-ping System. The data, archived in the geo-database, first used for 3D building model definition and street network model, have also been used for direct vulnerability as-sessment of buildings and street networks and for the analysis of mutual indirect vul-nerability induced by seismic hazard.

In this paper, results of the vulnerability assessment carried out on Reinforced Concrete buildings are presented. Such assessment is based on an analytical method in which a simulated design procedure is employed to evaluate the building structural characteristics based on few input data, and the seismic capacity is evaluated through the N2 method, leading to the construction of fragility curves. Hence, based on the specific seismic hazard evaluated for the case study of the Municipality of Avellino, failure probabilities in a given time window and for given building performance lev-els can be calculated. Different data sources – namely the filed survey and the air-borne LIDAR data, characterized by different detail level and time demand – can be assumed to define the input data to this procedure; in this way, a multilevel seismic vulnerability assessment is carried out, and results are compared.

2 Avellino City

The case study here described has been conducted in the area of the Municipality of Avellino, in the Campania region (Italy). Avellino (40°54′55″N 14°47′23″E, 348 m a.s.l., 42 km NE of Naples, Total population: 52,700) is situated in a plain called “Conca di Avellino” and surrounded by mountains: Massiccio del Partenio (Monti di Avella, Montevergine and Pizzo d’Alvano) on NO and Monti Picentini on SE. Due to the Highway A16 and to other major roads, Avellino also represents an important hub on the road from Salerno to Benevento and from Naples.

Avellino was struck strongly by the disastrous Irpinia earthquake of 23 November 1980. Measuring 6.89 on the Richter Scale, the quake killed 2,914 people, injured more than 80,000 and left 280,000 homeless. Towns in the province of Avellino were hardest hit and the Italian Government spent during the last thirty years around 30 billions of Euro on reconstruction. Consequently to the earthquake and to regulate the reconstruction activities, several specific acts, decrees, zoning laws and ordinance have been issued: the first one was the Law n. 219/1981 [1], that entrusted the urban planning to the damaged Municipalities, under the coordination of the Campania Region. In the same year, Avellino was also classified for the first time as seismic in technical building code [2]. From 2006 the urban planning issues of Avellino and neighbour areas are regulated by two instruments: P.I.C.A. (Italian acronym that stands for Integrated Project for Avellino City) and P.U.C. (Urban Plan for Avellino Municipality).

Seismic Vulnerability Assessment Using Field Survey and Remote Sensing Techniques 111

3 Seismic Hazard

In the present paper, seismic hazard is evaluated according to the Italian National Technical Standards [3]. According to these standards, the seismic hazard is defined in terms of the maximum horizontal expected acceleration in free field conditions on a rigid site of reference with horizontal topographic surface, and in terms of the elastic acceleration response spectrum, with reference to pre-defined exceeding probability over the reference period. The spectral shapes are defined according to the exceeding probability during the reference period (PVR), from the following parameters values referred to rigid horizontal site:

• ag maximum acceleration at the site; • F0 maximum value of amplification factor of the spectrum in horizontal

acceleration; • TC

* value of the initial period corresponding to the constant speed branch of the spectrum.

These parameters, which are tabulated in Annex B of NTC, are provided by [4] in correspondence to the points of a grid whose nodes are sufficiently close together (not more than 10 km away), and for different return periods TR, that fall into a target range between 30 and 2475 years, extremes included. For intermediate values of seismic intensity, an interpolation procedure is proposed. Nevertheless, in this study there is the need to extend elastic demand spectra above and below the ex-treme values. To this aim, the formulations proposed for the interpolation procedure are also used to extrapolate the above mentioned parameters out of the given range of values.

Fig. 1. Stratigraphic and topographic soil conditions in Avellino city

112 P. Ricci et al.

The city of Avellino is between four points belonging to the grid, therefore in order to derive the values of the parameters ag, F0 and TC

* it is necessary to apply a method of interpolation based on the distance from the site of interest to each of the four grid points.

In this way, seismic hazard expressed in terms of maximum effective ground ac-celeration and elastic response spectrum at the bedrock is determined, but these values should be calculated taking into account the different categories of soil that character-ize a site, because the land surface exerts a filtering function and tends to vary the frequency content of the seismic signal. According to the NTC, to consider the strati-graphic amplification it is sufficient to consider the last 30 m of ground from the foundation; hence, five soil categories (from A to E) are defined based on the average speed of propagation of shear waves into the ground in the last 30 m (Vs,30). Such information has been determined from microzonation data. Then, stratigraphic effects are taken into account through coefficients depending on the soil category of the site of interest. Effects related to the topographic conditions are considered, too, by means of an amplifying coefficient depending on the topographic category (from T1 to T4), which is determined by spatial processing the DTM of the city in order to obtain the slope surface at any point.

4 Field Survey on Building Stock

Building stock data have been collected through a detailed field survey carried out by means of a survey form implemented in a tablet PC, see Fig. 2. In this Section, the form prepared to carry out the field survey on building stock is presented first. Hence, the analysis of collected data is presented, particularly focused on RC buildings.

4.1 Survey Form

Survey form was sub-divided in different sections:

• Section 1 – Building identification, reporting data for the identification of the building, consisting of the address, the assigned ID code and data concerning the execution of the survey. Moreover, the building position was identified as “iso-lated” or “adjacent” to other buildings. In the latter case, the position of the build-ing with respect to the adjacent ones was reported (namely as internal, external or corner). A schematic representation of building plan view with the measured di-mensions was reported, too.

• Section 2 – Building description, reporting main data necessary for describing the building were reported, i.e. the number of storeys (above and below the ground level), the average surface area, the age of construction and the main destination use. The latter may represent an useful data in view of a loos estimation procedure.

• Section 3 – Building morphology, reporting a detailed representation of morphol-ogic building characteristics. The plan shape was identified first, referring to six main typologies. Then, height dimensions were reported, also considering a possi-ble difference between the 1st storey height and the upper storey height. Finally, data about the regularity in elevation were reported, including the number of sto-reys characterized by a reduction in surface area (if present) and the ground slope.


• Section 4 – Structural typology, reporting the identification of the structural typol-ogy, namely reinforced concrete, masonry, steel or mixed typology. In the case of reinforced concrete, a further indication was provided, aimed at identifying the specific structural system, i.e. frame, shear wall or dual system.

• Section 5 – Infills, reporting data about infill distribution. In particular, the ar-rangement of infill panels at the bottom storey was described. Infill typology was also reported.

• Section 6 – Bays, reporting the number of bays per each side (according to the plan shape).

• Section 7 – Stairs presence and position, reporting the number and the position (central or lateral) of staircases per each side of the building.

It is to be noted that some of the illustrated sections (e.g., infill description) spe-cifically fit reinforced concrete structural typology. The illustrated survey form in-cludes the main parameters – among the ones that can be reasonably collected during a field survey – that may have a significant influence on building seismic capacity, addressing a particular attention to specific potential sources of seismic vulnerability.

Fig. 2. Example screenshots of tablet software used in the field survey

4.2 Analysis of Building Stock Data

On the whole, 1327 buildings were surveyed. Among these, 1058 were RC buildings and 265 were masonry buildings, see Fig. 3a. Steel and mixed buildings were present in negligible percentages (only 4 buildings out of 1327).

As far as the age of construction is concerned, in 254 buildings (almost 20% of the population) it was not determined. However, a large part of masonry building stock was constructed at the beginning of the 20th century or before, or early after World War II. The period characterized by the highest growth of RC building stock is around 1970s and 1980s, see Fig. 3b. It is noted that a quite large part of the RC building stock was constructed after the 1980 earthquake. Pre- and post-81 buildings respec-tively represent about the 56 and 44% of the RC building population whose age of construction was determined.

114 P. Ricci et al.

Interesting observations can be made looking at data concerning the distribution of infills. The opening percentage in infill panels at the bottom storeys of rectangular RC buildings is shown in Fig. 3d. This percentage was evaluated as the weighted average of the opening percentage on the four sides of the building. A peculiar trend is ob-served, characterized by two intermediate “peaks” corresponding to about 25 and 50%, due to the relatively high frequency of buildings fully infilled on two or three sides and without infills on the remaining. It is to be noted that a 100% opening per-centage corresponds to Pilotis buildings.

(a) (b)

(c) (d)

Fig. 3. (a) Structural typology of surveyed buildings; (b) age of construction of RC buildings; (c) number of storeys of RC buildings; (d) infill opening percentage at bottom storey in RC buildings

5 Remote Sensing (RS) Data and Techniques

Remote Sensing (RS) data and techniques are the main source of a wide range of environmental information about landscape and its characteristics, especially in the case of studies concerning urbanized areas. Moreover, RS procedures for urban map-ping and monitoring offer significant advantages in terms of cost effectiveness and timeliness in the availability of information over larger areas. Therefore, RS data, opportunely processed and elaborated, can give a strong support in monitoring tasks and are essential for an effective and sustainable urban planning and management [5].

In this general framework, gathering information about buildings 3D geometry (height and/or number of storeys) is fundamental for extensively evaluating the vulner-ability within the study area. The recent progress of RS techniques (sensors, platforms and systems) makes them very attractive and capable to contribute to the above men-tioned monitoring tasks in terms of geometric resolution, operability and accuracy. Thus, a specific methodology has been implemented and calibrated to extract 3D build-ings parameters using RS data acquired from aerospatial platforms, in particular by


means of active LIDAR (Light Detection and Ranging) technology, which allowed to assess the height and planimetric shape of buildings.

Nowadays, LIDAR has emerged as an effective technology for the acquisition of high quality Digital Surface Models (DSM), due to its ability to generate 3D dense terrain point cloud data with high accuracy [6]. Recent advances of airborne laser scanning systems and techniques has opened a new way of directly measuring eleva-tion or generating DSM with high vertical and temporal accuracies. Compared with traditional photogrammetry, LIDAR has advantages in measuring 3D objects in terms of accuracy and density, automation, and fast delivery time [7].

Referring to the case-study here described, a LIDAR airborne RS mission has been planned and carried out in 2007 over the entire municipality of Avellino, using an Optech ALTM 3100 system and acquiring range point clouds data with a density of 4 points for square meter. Moreover on the same area a 31 band hyperspectral data acquisition has been carried out using the AISA-Eagle programmable sensor on board of the same airborne platform.

The 3D geometric parameters of buildings were extensively obtained through a methodology integrating active LIDAR technology [8] [9], aerophotogrammetry [10] and GIS techniques [11]. The exploitation of GIS and active/passive RS techniques coupled with tailored ground calibrations of above described procedures has allowed a detailed estimation of geometric and typological attributes for each building in test areas, in order to support the vulnerability assessment.

5.1 LIDAR Data Processing

The LIDAR raw data, once preliminary filtered and georeferenced, has been proc-essed through customised methodologies, implemented and calibrated on purpose for the area of interest [12], in order to firstly extract the DSM (Fig. 4a). That model has been initially used to properly identify the ground points for extracting the DTM (Digital Terrain Model, Fig. 4b) through interpolation; then, manmade objects like buildings have been extracted from non-ground points, so that it has been possible to derive extensively height values for each building within test area [13]. In particular, the processing schema followed to extract building features and vegetation attributes is reported in Fig. 5. As you can see the first step was the bare earth production via multistep morphological and dimensional filtering of the LIDAR multiple returns data and taking into account also the typical walls slope range of manmade objects (essentially buildings) raising from the mean ground level that must be throw out. In the subsequent building extraction step, were considered minimum/maximum area and elevation values; further, textures parameters were also included for vegetation characterization. Of course, all these parameters needed to be tuned over the entire area of study: to obtain suitable results, this step was accomplished by preventive evaluation of output obtained on different test sites, representative of the main urban typologies. Given that the Optech ALTM 3100 is able to handle the laser multi-returns from semitransparent tree canopies, more reliable measurements has been carried out also in those zones where it was difficult to outline the shape of the buildings because of their partially occlusion by trees or vegetation. In addition, exploiting the capability of LIDAR data to offer a better planimetric and altimetric description of the buildings, it has also been possible to obtain the geometric characterization of buildings roofs (in terms of shape) and the location and distribution of trees in the study areas.

116 P. Ricci et al.

Fig. 4. DSM (a) and DTM (b) extracted from LIDAR (in grey tones the elevation values)

Fig. 5. Building feature extraction from LIDAR data: processing schema

5.2 GIS Procedures

Subsequently, GIS procedures have been specifically implemented to describe the built-up areas and to store all the data acquired and the information produced in a specific GeoDatabase. This former has been suitably designed to be included in DSS tools/procedures devoted to support planning and decision making in case of different risk scenarios. The digital cartography at 1:2,000 scale of the Municipality of Avellino


has been used to overlay vector information about the buildings (especially, their foot-print) with LIDAR data [14]. Therefore, combining digital cartography and height val-ues coming from LIDAR, for each building geometric attributes and morphological features have been extracted in a semi-automatic way: area, perimeter, volume, total height of the building and ground altitude beneath itself. In particular, this task has been carried out using the following approach [15]:

• buildings footprints in shapefile vector format (extracted from 1:2,000 Cartogra-phy) have been overlaid to LIDAR data;

• by means of GIS zonal algorithm, to each feature object have been automatically assigned the respective geometric (3D) and morphological attributes.

Fig. 6. Buildings footprints and LIDAR DSM overlay: on the left, attribute table of the build-ings shapefile enhanced with the 3D information extracted from LIDAR

Following this approach, geometric accuracy and topological rules typical of Cartography have been preserved and, at the same time, an enhanced version of buildings shapefile has been produced, thanks to information extracted from LI-DAR data and stored as new attributes of its polygonal elements (buildings). The updated version of Cartography constituted the basis of the GeoDatabase. More-over, this approach makes also possible to easily update existing digital maps, using the 3D data of buildings obtained by means the above mentioned methodology [16]. Finally, the updated Cartography has been overlaid with ISTAT [17] Census data, in order to extract information about buildings typology and construction age, ag-gregated for census section.

118 P. Ricci et al.

6 Seismic Vulnerability Assessment

6.1 Procedure

Methodologies for the assessment of seismic vulnerability of buildings provide the probability of a given level of damage as a function of a parameter representing the seismic intensity. Generally speaking, seismic vulnerability can be assessed by means of empirical or analytical methodologies: in the first case the assessment of the ex-pected damage for a given building typology is based on the observation of damage suffered during past seismic events; in the latter case the relationship between seismic intensity and expected damage is provided by a model with direct physical meaning (e.g., [18] [19]).

In this work, a simplified analytical method for the evaluation of seismic vulner-ability of Reinforced Concrete (RC) buildings has been applied [20]; this method is based on a simulated design procedure [21] to evaluate the building structural charac-teristics based on few data such as number of storeys, global dimensions and type of design, and on the simplified assumption of a Shear Type behaviour to evaluate in closed form the lateral non-linear static force-displacement response. The structural model also includes elements representing the infill panels, which are usually consid-ered as non-structural elements but can highly influence the seismic response of RC buildings. The assessment of the seismic capacity is carried out within the framework of the N2 method [22], leading to the construction of fragility curves, which represent a relationship between a seismic intensity parameter and the corresponding exceeding probability of a given damage threshold (typically represented by a displacement capacity). Finally, the failure probability in a given time window and for given Limit States can be calculated, based on the seismic hazard defined in Section 3. Possible future developments of this procedure may include the consideration of brittle failure mechanisms due to the interaction between structural and non-structural (infill) ele-ments and structural irregularity in plan and/or in elevation. Results of the application of such simplified procedure will be shown in next Section. Input data, for each build-ing, are represented by:

• global geometrical parameters such as number of storeys, height and plan dimensions;

• distribution of infill panels; • type of design and values of allowable material stresses employed in the

simulated design procedure; • material characteristics.

Field survey (see Section 4) provides data about global building dimensions and distribution of infill panels, besides the age of construction. Based on the latter infor-mation, the type of design (seismic or gravitational) is determined, since Avellino was classified as seismic for the first time in 1981 [2], following the 1980 Irpinia earth-quake. Moreover, material characteristics employed in the determination of the struc-tural response – such as concrete compressive strength, steel yield strength and type of reinforcement – and material allowable stresses employed in the simulated design procedure are determined from the age of construction, too, based on statistical stud-ies on mechanical characteristics of these materials and on past code prescriptions.


6.2 Results Based on Field Survey Data

Results of the seismic vulnerability assessment carried out according to the previously illustrated procedure can be reported in terms of failure probability at different Limit States in a time window of 1 year, obtained by convolving the fragility curves of each single building with the hazard curve representing the exceeding probability of a seismic intensity parameter (i.e., the Peak Ground Acceleration, PGA) in the chosen time window. Two Limit States are considered: Damage Limitation (DL) and Near Collapse (NC). First, evaluated failure probabilities at DL and NC Limit States can be reported as a function of the number of storeys (see Fig. 7). Mean values of Pf are reported as red squares. A clear trend is observed, showing in both cases the higher vulnerability of taller buildings. Failure probabilities at DL Limit State are lower by an order of magnitude, on average, with respect to the corresponding probabilities at NC. The increase in Pf with the building height is not uniform, rather highlighting a sudden increase when the number of storeys changes from the range of values Nstoreys ≤ 3 to Nstoreys ≥ 4. The influence of the age of construction can be observed by report-ing the failure probabilities, at both Limit States, for pre-1981 and post-1981 build-ings. As a matter of fact, the major change represented by the introduction of seismic load prescriptions in Avellino city took place in 1981.

Fig. 7. Annual failure probability depending on the number of storeys at DL (a) and NC (b) Limit States

Fig. 8 reports mean failure probabilities at DL and NC Limit States in pre- and post-1981 buildings. If a relative comparison is carried out for equal values of Nstoreys, a higher vulnerability in pre-1981 buildings, as expected, can be generally observed, particularly at NC Limit State. However, most clear is the dependence on the number of storeys. Seismic vulnerability can also be analyzed by observing the spatial distri-bution of failure probability. In particular, the spatial distribution can be referred to single census cells reporting the corresponding average values among all buildings in the census cell.

Fig. 9-a reports the average annual failure probability at NC Limit State per census cell; higher values are observed in central and North-Western areas. A clear influence of the difference in seismic hazard due to a different soil type (see Section 3) can be recognized, see Fig. 9-b, leading, as expected, to higher failure probabilities, on aver-age, for buildings located on soil type E.

120 P. Ricci et al.

Fig. 8. Comparison between annual failure probabilities at DL (a) and NC (b) Limit States in pre- and post-1981 buildings

Fig. 9. (a) Average annual failure probability of rectangular RC buildings at NC and (b) most widespread soil type per census cell

6.3 Results Based on LIDAR Data

The procedure described in Section 6.1 can also be carried out assuming for the geo-metric building parameters the LIDAR data (Section 5) instead of the data from the field survey (Section 4). In this way, the effectiveness of a seismic vulnerability as-sessment based on a different data source can be investigated, according to a multi-level approach. Hence, the sources of information about building characteristics for the application of the seismic vulnerability assessment procedure are:

• LIDAR data, providing (based also on cartography) global dimensions of buildings;

• 2001 ISTAT [17] census data, providing, although in aggregated form, infor-mation about the structural typology (masonry or RC) and the age of construc-tion of buildings in each census cell;

• Statistics on interstorey height (both at first and upper storeys), bay length and infill distribution from the field survey described in Section 4, which may be considered as literature data to be used as “prior” information on RC building stock characteristics for any Italian city.


Based on these data, the following parameters are determined:

• Age of construction, assumed as a Random Variable characterized by a dis-crete probability distribution evaluated from census data within each census cell. If both masonry and RC buildings are present in the same census cell, such distribution is evaluated assuming that all masonry buildings are older than RC buildings. This assumption is needed because data are provided in ag-gregated form.

• Infill opening percentage at bottom storey, assumed as a Random Variable characterized by a simplified discrete probability distribution evaluated from field survey data; based on these data, 0, 25 and 50% are assumed as equally probable values.

• Number of storeys and number of bays (in both building directions), respec-tively based on height and plan dimensions. In order to evaluate these parame-ters, interstorey height at first and upper storeys and bay length are calculated as the values providing the least scatter with median values provided by field survey (i.e., 3.3 m for first storey height, 3.2 m for upper storey height and 4.4 m for bay length).

It is noted that the remaining parameters needed as input data to the seismic vulner-ability assessment procedure – namely the type of design, material characteristics (both median values and corresponding variability) and material allowable stresses employed in the simulated design procedure – are evaluated from the age of construction. An example application of the procedure based on LIDAR data is carried out on 7 adjacent census cells in south-eastern city area, including 59 rectangular RC buildings.

In the selected area, compared with remaining city areas, a good agreement be-tween LIDAR and field survey data is shown. Nevertheless, compared with data pro-vided by field survey, LIDAR data overestimate building height, on average, by about 1.1 m. A reason for this bias comes from the presence of inclined roofs, leading to an estimate of mean building height (within the polygonal shape provided by cartogra-phy) generally higher compared with the survey measure, which does not account for the roof. A wrong estimate of building height can lead to an error in the evaluation of the number of storeys, compared with field survey data: among the 59 considered buildings, in 3 cases the estimated Nstoreys is lower and in 37 cases greater than the actual value. However, the error in the evaluation of Nstoreys is never higher than 1. Hence, the failure probabilities obtained with different data sources (field survey or LIDAR) can be compared, evaluating the difference between the latter and the for-mer; such difference can be considered the error resulting from an application of the procedure based on more poor data. As expected, it is found that an overestimate of Nstoreys leads, on average, to an overestimate of Pf,NC. The actual infill opening per-centage at bottom storey, provided by field survey, also influences the error in esti-mating Pf,NC when LIDAR data are used. As a matter of fact, Pf,NC increases as the infill opening percentage increases; therefore, the failure probability based on LIDAR data overestimates the corresponding value based on field survey data for low values of infill opening percentage, and vice. Nevertheless, even if the actual infill opening percentage is equal to the central value of the uniform discrete probability distribution assumed for this Variable when using LIDAR data (i.e., 25%), an overestimate of

122 P. Ricci et al.

Pf,NC is observed, independent of the possible error in Nstoreys. This is due to the fact that the increase in Pf,NC when the infill opening percentage increases is greater than the corresponding opposite decrease. Therefore, a lack of knowledge about this pa-rameter necessarily leads to an overestimate of the failure probability.

However, a comparison between results based on different data sources should be more properly carried out considering census cells instead of single buildings, consis-tent with the scale of the seismic vulnerability assessment. Fig. 10 shows the average error in Pf,NC within the analyzed census cells: a generally acceptable scatter is observed. Moreover, it is to be noted that, from a qualitative standpoint, a risk overes-timation has to be expected when data employed in the assessment procedure are characterized by a lower knowledge level, that is, by a higher uncertainty.

Fig. 10. Average error in Pf,NC per census cell

7 Discussion and Results

The preliminary assessment of buildings vulnerability in the selected test areas has been possible using the extensive estimation of their geometrical parameter (stored in the above mentioned GeoDatabase) and considering the typology data coming from in situ campaigns. A further improvement could be attempted to reduce and optimise the expensive ground. Considering that in the same LIDAR flight, semi-photogrammetric RGB and AISA-Eagle hyperspectral strips has been also taken, in the immediate fu-ture the assessment of buildings vulnerability could receive a strong effort from the combination of other advanced techniques (clustering analysis, data mining,…), to be implemented using all the above mentioned data. In this perspective, the cost/effectiveness of the implemented methodology would be increased, reducing the costs of the ground campaigns and the production time thanks to the automation and the accuracy of the new process. The cost of airborne LIDAR survey on Avellino municipality has amounted at about 1000 €€ /Km2, including the preprocessing task for georeferencing, prefiltering and orthocorrection and the color airphotos acquisition. In general, this price proves to be cost-effective if compared with the cost of similar services currently available, also considering the fact that, as often as not, for the same amount are included only LIDAR preprocessed data (at lower ground resolu-tion). In the last years, many European Central and Local Administrations have started off extensive LIDAR acquisition programs, like the Italian Ministry of Envi-ronment. It is desirable that those data will be made widely available to operators and scientific community, to implement methods able to support a strategic territorial


planning. In order to overcome the lack of such information about other areas vulner-able to earthquakes or natural disasters, the most recent satellite platforms (high resolution active, like SAR or Cosmo-Skymed and/or passive like QuickBird and WorldView) make available a large amount of data that will be processed to assess 3D information at building level of detail, although further R&S work is still needed in this framework.

A seismic vulnerability assessment was carried out on surveyed Reinforced Con-crete buildings by means of an analytical methodology. Such procedure was first applied based on the information gathered from the field survey, providing data such as detailed geometric characteristics, structural typology and age of construction of each single building, thus leading to results assumed as a reference due to the high detail level and reliability of the input data. An analysis of these results highlighted that the seismic vulnerability was mainly influenced by the age of construction and, more important, by the number of storeys and the soil stratigraphic condition.

LIDAR data, combined with ISTAT census data, were then used as an alternative source of information for the same procedure, on a selected test area, in order to in-vestigate the approximation in the seismic vulnerability assessment due to the use of less detailed data. A comparison between the results obtained with different sources of information highlighted, in some cases, a risk overestimate when less detailed data were used, i.e. LIDAR instead of field survey. Such overestimate, in average terms, is also consistent with the higher uncertainty affecting an assessment procedure based on more poor data. However, a generally acceptable scatter was observed, thus con-firming the importance to make further efforts in the investigation and the develop-ment of relatively cheap, fast and easy to apply alternative techniques based on Re-mote Sensing for the acquisition of data on existing building stock to be used in large scale seismic vulnerability assessment procedures.

References

1. Legge n. 219 del 14/5/1981: Conversione in legge, con modificazioni, del decreto legge 19 marzo 1981, n. 75, recante ulteriori interventi in favore delle popolazioni colpite dagli eventi sismici del novembre 1980 e del febbraio 1981. Provvedimenti organici per la ricos-truzione e lo sviluppo dei territori colpiti. G.U. n. 134 del 18/5/1981 (in Italian)

2. Decreto Ministeriale del 7/3/1981: Dichiarazione di zone sismiche nelle regioni Basilicata, Campania e Puglia (in Italian)

3. Decreto Ministeriale del 14/1/2008: Approvazione delle nuove norme tecniche per le costruzioni. G.U. n. 29 del 4/2/2008 (in Italian)

4. INGV-DPC S1: Progetto S1. Proseguimento della assistenza al DPC per il completamento e la gestione della mappa di pericolosità sismica prevista dall’Ordinanza PCM 3274 e pro-gettazione di ulteriori sviluppi. Istituto Nazionale di Geofisica e Vulcanologia – Diparti-mento della Protezione Civile (2007) (in Italian), http://esse1.mi.ingv.it

5. Forster, B.C.: An examination of some problems and solutions in monitoring urban areas from satellite platforms. Int. J. Remote Sens. 6(1), 139–151 (1985)

6. Ackermann, F.: Airborne Laser Scanning for Elevation Models. GIM - Geomatics Info Magazine 10, 24–25 (1996)

7. Baltsavias, E.: A comparison between photogrammetry and laser scanning. ISPRS J. Pho-togramm. 54, 83–94 (1999)

124 P. Ricci et al.

8. Axelsson, P.: Processing of laser scanner data: algorithms and applications. ISPRS J. Pho-togramm. 54, 138–147 (1999)

9. Crosilla, F., Visintini, D., Prearo, G., Fico, B.: Esperienze di filtraggio, classificazione segmentazione e modellazione di dati spaziali da rilievo laser aereo. Bollettino SIFET 1(05), 13–51 (2007)

10. Gabet, L., Giraudon, G., Renouard, L.: Automatic generation of high resolution urban zone digital elevation models. ISPRS J. Photogramm. 52, 33–47 (1996)

11. Haala, N., Brenner, C., Anders, K.-H.: Generation of 3D city models from digital surface models and 2D GIS. International Archives of Photogrammetry and Remote Sensing 32(3-4W2), 68–75 (1997)

12. Ma, R.: DEM Generation and Building Detection from Lidar Data. Photogramm. Eng. Rem. S. 71(7), 847–854 (2005)

13. Sohn, G., Dowman, I.J.: Data fusion of high-resolution satellite imagery and LiDAR data for automatic building extraction. ISPRS J. Photogramm. 62, 43–63 (2007)

14. Carneiro, C., Silva, V., Golay, F.: Incorporation of Morphological Properties of Buildings” Descriptors Computed from GIS and LIDAR Data on an Urban Multi-agent Vector Based Geo-simulator. In: Gervasi, O., Taniar, D., Murgante, B., Laganà, A., Mun, Y., Gavrilova, M.L. (eds.) ICCSA 2009. LNCS, vol. 5592, pp. 205–220. Springer, Heidelberg (2009)

15. Borfecchia, F., De Cecco, L., Pollino, M., La Porta, L., Lugari, A., Martini, S., Ristoratore, E., Pascale, C.: Active and passive remote sensing for supporting the evaluation of the ur-ban seismic vulnerability. Italian Journal of Remote Sensing 42(3), 129–141 (2010)

16. Awrangjeb, M., Ravanbakhsh, M., Fraser, C.S.: Automatic detection of residential build-ings using LIDAR data and multispectral imagery. ISPRS J. Photogramm. 65, 457–467 (2010)

17. ISTAT - Italian Institute of Statistics, http://www.istat.it 18. Cosenza, E., Manfredi, G., Polese, M., Verderame, G.M.: A multilevel approach to the ca-

pacity assessment of existing RC buildings. Journal of Earthquake Engineering 9(1), 1–22 (2005)

19. Polese, M., Verderame, G.M., Mariniello, C., Iervolino, I., Manfredi, G.: Vulnerability analysis for gravity load designed RC buildings in Naples – Italy. Journal of Earthquake Engineering 12(S2), 234–245 (2008)

20. Ricci, P.: Seismic vulnerability of existing RC buildings. PhD thesis, University of Naples Federico II, Naples, Italy (2010)

21. Verderame, G.M., Polese, M., Mariniello, C., Manfredi, G.: A simulated design procedure for the assessment of seismic capacity of existing reinforced concrete buildings. Advances in Engineering Software 41(2), 323–335 (2010)

22. Fajfar, P.: Capacity spectrum method based on inelastic demand spectra. Earthquake Engi-neering and Structural Dynamics 28(9), 979–993 (1999)

Seismic vulnerability assessment using field survey and remote sensing techniques

Documents