THE USE OF GIS FOR THE APPLICATION OF THE PHENOMENOLOGICAL APPROACH TO THE SEISMIC RISK ANALYSIS : THE CASE OF THE ITALIAN FORTIFIED ARCHITECTURE

The present paper proposes the use of GIS for the application of the so-called phenomenological approach to the analysis of the seismic behaviour of historical buildings. This approach is based on the awareness that the different masonry building typologies are characterized by different, recurring vulnerabilities. Thus, the observation and classification of the real damage is seen as the first step for recognizing and classifying these vulnerabilities, in order to plan focused preventive interventions. For these purposes, the GIS has proven to be a powerful instrument to collect and manage this type of information on a large number of cases. This paper specifically focuses on the application of the phenomenological approach to the analysis of the seismic behaviour of fortified buildings, including castles, fortresses, citadels, and all the typical historical constructions characterized by the presence of massive towers and defensive walls. The main earthquakes which struck Italy in the last 40 years (up to the recent Central Italy seismic swarm) were taken into consideration and described by means of shake maps. A previously published work has been continued with the addition of new data and some improvements, including a specific symbology for the description of building typologies and conservation status on the maps, the indications of damage levels and the comparison between shake maps in terms of pga and in terms of pseudo-acceleration. The increase in knowledge obtained and the broader frame given by the analysis of the data are here directed to the primary aim of cultural heritage preservation. * Corresponding author


INTRODUCTION 1.1 The GIS for cultural heritage management
The use of Geographic Information System (GIS) for the archival, analysis and representation of geographic information has become a consolidated tool in different scientific fields.Also in the field of architectural heritage, the GIS technology is used to join and relate data coming from different disciplines involved in documentation and maintenance processes of Cultural Heritage assets(MIBACT2015).In Italy the Ministry of Cultural Heritage has introduced a GIS specifically for the risk management of cultural heritage (MIBACT 2016), but it is more focused on the definition of seismic and geo-morphological hazard levels than to the definition of the vulnerability, which is only related to the conservation status; moreover, no specific data are reported on the damages suffered in previous earthquakes.Some GIS have been applied specifically to fortified buildings: Monti (2006) on the Emilia-Romagna castles and Deidda et al. (2015) for the Sardinia coastal towers, among others, but they have no specific reference to the seismic issue or to the damage they show.The first work that applied GIS for the seismic damage and behaviour of fortifies architecture is Coisson et al (2016).The present paper continues this work with more data and some improvements.In particular, the last Central Italy seismic swarm, up to the January 2017 events, has been included with specific on site surveys, and the number of castles overall considered has increased, with a better selection of the case studies.

The phenomenological approach in the seismic field
In particular, the present paper proposes the use of GIS for the application of the so-called phenomenological approach to the analysis of the seismic behaviour of historical buildings.This approach is based on the awareness that the different masonry building typologies are characterized by different, recurring vulnerabilities.Thus, the observation and classification of the real damage is seen as the first step for a focused preventive intervention.This approach was first applied on a systematic base in Italy to the study of the effects of the Friuli 1976 earthquake on churches (Doglioni et al., 1994).This work later became the base for the introduction of the table of recurring damage mechanisms for churches in the latest Italian technical code for the protection of cultural heritage from the seismic risk (Guidelines 2007), recognizing its validity.The GIS is here proposed as a powerful instrument to collect and manage this type of information on a large number of cases, increasing the effectiveness of the approach.The present work specifically focuses on the application of the phenomenological approach to the analysis of the seismic behaviour of fortified buildings, including castles, fortresses, citadels, and all the typical historical constructions characterized by the presence of towers and defensive walls.These buildings, despite being a meaningful element of the architectural landscape, have been subjected to limited studies compared to other building typologies, as far as the specific seismic damages are concerned, and mostly regarding single case studies (Tiberti et al. 2016, Castellazzi et al. 2015, Coïsson et al. 2016a, Meschini et al 2015, Barrile et al. 2016, Leoni et al. 2015).Only since the 2012 Emilia earthquake, a more systematic approach was applied with the definition of a table for the recurring damage mechanisms specific for the fortified buildings damaged in that occasion (Cattari et al, 2012 andCattari et al., 2014).The validity of the table was then widened by the following work of Coisson et al (2016), which is at the base of the present paper.

The fortified architecture typologies
Fortified architecture is a generic term, which can indicate a very wide variety of buildings, usually characterized by the presence of towers and defensive walls, but with features varying depending on the geographic area, on the materials availability, on the history, on the importance and specific use of the construction.In an attempt of cataloguing these possible variations, Perogalli (Fig. 1) has identified 16 different types of fortified buildings, form castles to fortified bridges, and proposed specific symbols to represent them in maps.Furthermore, to represent the consistency, he proposed to plot them with a black dotted line (historical traces), black line with white hatch (ruins) or black hatch (entire).He also proposed an additional symbol, to be designed below the typological symbol, to describe the maintenance state.For our purposes, the historical traces were not taken into consideration and the two parameters of consistency and maintenance were merged into the global concept of state of conservation, with a specific symbology (Fig. 2) which derives from the ones proposed by Perogalli.Indeed, being the vulnerability to seismic damage the main objective of the research, a good maintenance in a ruined castle has not the same implications as a good maintenance in an entire fortress.Starting from Perogalli's proposal, thus, the symbols have been modified as indicated in Figure 2 and have been inserted in the GIS software.Moreover, in case of buildings damaged by the earthquake, the black colour has been replaced with a white one.Figure 3 shows the use of the typology symbols for the Central Italy earthquakes whereas Figure 3 shows the state of conservation for the same fortified buildings.

THE PROPOSED APPROACH
The main seismic events which struck Italy in the last 40 years have been analysed in Coisson et al. (2016).In this paper the analysis is widened (Fig. 4) with the inclusion of the recent Central Italy earthquake 2016-17, that has damaged a large number of buildings, some of which were already affected by the Umbria-Marche earthquake of 1997.For instance, a list of the seismic events considered for Central Italy, with date, coordinates of the epicentre and magnitude is shown in Table 1.
For each seismic event, the characteristics of the earthquake have been described by means of shake maps.

Shake Maps
A ShakeMap is a representation of ground shaking produced by an earthquake.The information it presents is different from the earthquake magnitude and epicenter that are released after an earthquake because ShakeMap focuses on the ground shaking produced by the earthquake, rather than the parameters describing the earthquake source.So, while an earthquake has one magnitude and one epicenter, it produces a range of ground shaking levels at sites throughout the region depending on distance from the earthquake, the rock and soil conditions at sites, and variations in the propagation of seismic waves from the earthquake due to complexities in the structure of the Earth's crust.The shake maps can be generated not only in terms of peak ground acceleration, but also in terms of velocity, pseudoaccelerations or even an instrumentally-derived, estimated Modified Mercalli Intensity map.This map makes it easier to relate the recorded ground motions to the expected felt and damage distribution.The Instrumental Intensity map is based on a combined regression of recorded peak acceleration and velocity amplitudes.In the present paper, a comparison between the different shake maps in terms of effectiveness in representing the real damage on the fortified architecture is carried out.
With the current Italy station distribution, data gaps are common, particularly for smaller events and events near or outside the edge of the network.In order to stabilize contouring and minimize the misrepresentation of the ground motion pattern due to data gaps, some values were predicted in areas without data.Given the epicenter and magnitude, peak motion amplitudes in spare regions are estimated from the Joyner, Boore, and Fumal (1997), and Joyner and Boore (1988) attenuation curves.
In pga maps, the peak horizontal acceleration at each station is contoured in units of percent-g (where g = acceleration due to the force of gravity = 981 cm/s/s).The peak values of the vertical components are not used in the construction of the maps because they are, on average, lower than the horizontal amplitudes and ground motion prediction equations used to fill in data gaps between stations are based on peak horizontal amplitudes.The contour interval varies greatly and is based on the maximum recorded value over the network for each event.
For moderate to large events, the pattern of peak ground acceleration is typically quite complicated, with extreme variability over distances of a few km.This is attributed to the small scale geological differences near the sites that can significantly change the high-frequency acceleration amplitude and waveform character.Although distance to the causative fault clearly dominates the pattern, there are often exceptions, due to local focussing and amplification.This makes interpolation of ground motions at one site to a nearby neighbour somewhat risky.Peak acceleration pattern usually reflects what is felt from low levels of shaking up to moderate levels of damage.
For the present work, the shake maps were taken from the USGS archives (USGS 2016), except for Central Italy events, for which they have been loaded from the INGV archive (Shakemap Working Group 2016).Shake maps of two different variables have been considered: peak ground acceleration pga, and spectral acceleration psa.

The GIS
The geodatabase was created using the open source software QGIS version 2.12 (QGIS Development Team 2015).As ShakeMap processing does not occur in a Geographic Information System (GIS), a specific post-processing of the grid file into shapefiles was carried out for direct import into GIS.In this application, the shapefiles are contour polygons of the peak ground-motion amplitudes.These contour polygons are actually equal-valued donut-like polygons that sample the contour map at fine enough intervals to accurately represent the surface function (Fig. 5).The shapefiles independent of a GIS are generated using a shareware package which employs a 4-point method for contouring.

The damage data collection
More than 750 castles in the areas struck by the considered earthquakes have been identified and inserted into the specifically formulated GIS.Then, a thorough archival and in situ observation of the damages suffered by the most significant case histories was made, to collect the main information needed for the application of the empirical approach.Based upon a previously defined damage mechanisms table, represented in Fig. 8 for completeness, the type of damage suffered by each fortress in the GIS was analysed and classified, to make some considerations on the distribution of the different mechanisms on the territory, in relation with the different characteristics in terms of soil, materials, building typologies, and conservation status.For a more immediate visual observation of the relation between level of damage and seismic action, as previously indicated, a specific graphic markers for the level of damage and for building features (typology, conservation status, materials) was defined and adopted.For each earthquake, a table was filled in with all the affected fortified buildings, with province, identification number, and name.The type of masonry (solid clay bricks, stone units or rubble masonry, tuff units, mixed) and the geometry of the base of the towers (none, circular, rectangular, irregular) were also reported, together with the presence of merlons and protruding elements, such as pepper-pot turrets.The state of conservation at the time of the seism (ruin, used, abandoned) was reported.The value of pga max and I max obtained by querying the GIS shakemaps and the soil morphology (plain, soft slope, strong slope, and crest) were inserted.Lastly, the table was filled by numbers from 0 (no damage) to 5 (collapse), which represent the level of damage.

The data analysis
For the recent Central Italy seismic events, characterized by many subsequent shocks of large intensity, with epicentres several kilometres apart, a more in depth analysis was carried out.A description of the seismic event can be found in Michele et al., 2016.
Looking at the Figure 6, it is possible to observe that the area struck by the whole Central Italy sequence of earthquakes is vast and some fortresses that survived the first shock, have been damaged subsequently.For this reason, it was necessary to inspect, for each point of the map, which of the many seismic events occurred had produced the strongest effects.Therefore, the envelope of the seismic actions was calculated for all the considered area.In particular, the 9 shake maps of the events with magnitude I M ≥ 5 have been superimposed by means of the GIS software, thus obtaining the shake map of maximum pga represented in Figure 6.This first analysis on the peak ground acceleration maps showed some inconsistencies with the observed damage distribution levels (in the upper part of the map in Fig. 6).For this reason, also the maps of maximum pseudo-acceleration were computed both for structures with a first period of 0.3 s.As pointed out in Cattari (2014), the period of 0.3 is typical of massive masonry In this case, it is possible to observe a "peninsula" that reaches the upper left part of the map, explaining the damage suffered in this area (Fig. 7).Lastly, statistical analyses on these data have been conducted in order to inspect possible connections between building features, peak ground acceleration and damage mechanisms (Figure 9).Some of the analyses have already been reported in (Coisson et al 2016), indicating the relative frequency of the damage to the different elements and allowing to identify some threshold values for the damage expected.Moreover, the connection between the damage and the building features is evident, thus allowing to identify the most vulnerable elements on which the interventions should be focused.

CONCLUSIONS
The paper presents a GIS specifically formulated for the catalogue and analysis of the damage suffered by over 750 fortified architectures hit by large earthquakes in the last 40 years in Italy.For the graphic representation of the analysed buildings, also considering the large variability of the fortified constructions, a specific symbology was developed and applied, derived from the one proposed by Perogalli, Considering the large number of data to be not only collected but also compared, analysed and subjected to statistical processing, the adoption of a GIS turned out to be particularly suitable for the purpose, allowing to easily link the data about the damage suffered by each analysed building to its specific seismic input obtained through the elaboration of the ShakeMaps, given the strict connection to the geographic location.This approach, thanks to the direct geographic comparison, allowed to evidence the more strict connection of the damage suffered by the inspected building typologies with the pseudoacceleration with a period of 0,3 rather than with the pga, which is instead the parameter adopted by the Italian technical code for the assessment of the seismic vulnerability and the design of the strengthening.At last, it should be underlined that the proposed phenomenological approach applied to the fortified architecture, also thanks to the practical functionality of the GIS, supplies a sound tool not only to increase knowledge in this field, but also for the seismic protection of these buildings which characterize the Italian landscape, as it allows to identify the most vulnerable elements which require priority interventions, also in relation to the local expected seismic intensity.

Figure 4 .
Figure 4. Map of Italy with the investigated fortified buildings.The map is surrounded by the shakemaps (pga) of the main seismic events.

Figure 5 .
Figure 5. Shakemap (pga) of Central Italy (first seismic event in Tab. 1) with fortified buildings.The damaged buildings are white and epicentres are represented with stars.

Figure 6 .
Figure 6.Interpolation of the pga's maximum values of the seismic sequence that hit Central Italy since August 2016