Journal of Volcanology and Geothermal Research 276 (2014) 105–120

Contents lists available at ScienceDirect

Journal of Volcanology and Geothermal Research

journal homepage: www.elsevier.com/locate/jvolgeores

Volcanic risk assessment: Quantifying physical vulnerability in the

built environment
S.F. Jenkins a,b,⁎, R.J.S. Spence b, J.F.B.D. Fonseca c, R.U. Solidum d, T.M. Wilson e
a
Department of Earth Sciences, University of Bristol, Bristol BS8 1RJ, UK
b
Cambridge Architectural Research, Unit 6, 25 Gwydir Street, Cambridge CB1 2LG, UK
c
Physics Department, Instituto Superior Técnico, Av. Rovisco Pais, 1049-001 Lisbon, Portugal
d
Philippine Institute of Volcanology and Seismology, PHIVOLCS Building, C.P. Garcia Avenue, Diliman, Quezon City, 1101, Philippines
e
Department of Geological Sciences, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents structured and cost-effective methods for assessing the physical vulnerability of at-risk com-
Received 21 November 2013 munities to the range of volcanic hazards, developed as part of the MIA-VITA project (2009–2012). An initial as-
Accepted 2 March 2014 sessment of building and infrastructure vulnerability has been carried out for a set of broadly defined building
Available online 12 March 2014
types and infrastructure categories, with the likelihood of damage considered separately for projectile impact,
ash fall loading, pyroclastic density current dynamic pressure and earthquake ground shaking intensities. In re-
Keywords:
Volcanic risk assessment
fining these estimates for two case study areas: Kanlaon volcano in the Philippines and Fogo volcano in Cape
Volcanic hazards Verde, we have developed guidelines and methodologies for carrying out physical vulnerability assessments in
Eruption impacts the field. These include identifying primary building characteristics, such as construction material and method,
Building vulnerability as well as subsidiary characteristics, for example the size and prevalence of openings, that may be important in
Vulnerability curves assessing eruption impacts. At-risk buildings around Kanlaon were found to be dominated by timber frame build-
Infrastructure vulnerability ings that exhibit a high vulnerability to pyroclastic density currents, but a low vulnerability to failure from seismic
shaking. Around Fogo, the predominance of unreinforced masonry buildings with reinforced concrete slab roofs
suggests a high vulnerability to volcanic earthquake but a low vulnerability to ash fall loading. Given the impor-
tance of agriculture for local livelihoods around Kanlaon and Fogo, we discuss the potential impact of infrastruc-
ture vulnerability for local agricultural economies, with implications for volcanic areas worldwide. These
methodologies and tools go some way towards offering a standardised approach to carrying out future vulnera-
bility assessments for populated volcanic areas.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction a considerable extent upon the physical vulnerability of the different

Damaging volcanic processes vary widely in their potential impact
for the built environment over space and time. As a general rule, explo-
sive volcanic eruptions are more damaging and result in more wide-
spread impacts than their effusive counterparts. Immediate damage
includes violent destruction through lateral forces and vertical loads,
burial and exposure to high temperatures within pyroclastic density
currents, lahars and debris flows. Buildings or key components of infra-
structure that experience little or no physical damage can still be subject
to a reduced or lost functionality, often at relatively low hazard intensi-
ties. Longer-term impacts include a reduction in the health and socio-
economic wellbeing of affected communities through temporary or per-
manent relocation and a loss of housing or livelihood, particularly for
agricultural economies that rely on the land. These impacts depend to
⁎ Corresponding author at: Department of Earth Sciences, University of Bristol, Bristol
BS8 1RJ, UK. Tel.: +44 117 9545400; fax: +44 117 9253385.
E-mail address: Susanna.Jenkins@Bristol.ac.uk (S.F. Jenkins).
built components to the range of volcanic hazards. Physical vulnerabil-
ity assessments are thus vital in helping to forecast the range of damage
and disruption – and therefore casualties, losses and reconstruction
costs – that may result from a future eruption. They can also be used
to infer hazard dynamics, e.g. lateral dynamic pressures, from damage
assessments.
Data that inform physical vulnerability estimates are generally
sourced in three ways: 1) Collection of post-eruption empirical damage
data (e.g. Baxter et al., 2005; Wilson et al., 2011); 2) Experimental test-
ing of material and structural failure (e.g. Zuccaro, 2000); and 3) Theo-
retical calculations of material strengths (e.g. Petrazzuoli and Zuccaro,
2004; Jenkins et al., 2013). Quantitative observation data collected dur-
ing or immediately after a damaging event are scarce due to the danger
and inaccessibility of impacted zones and the comparatively low fre-
quency of large eruptions impacting urban areas. Experimental data
are also limited, and, where available, refer exclusively to materials,
strengths, building codes and standards exhibited by the tested building
stock. In the absence of observation data, theoretical calculations of

http://dx.doi.org/10.1016/j.jvolgeores.2014.03.002
0377-0273/© 2014 Elsevier B.V. All rights reserved.
106 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120
building response to hazard parameters add greatly to the availability of
quantitative information and, often in conjunction with engineering ex-
pert judgement, can support the creation of vulnerability functions. Vul-
nerability functions outline the probability of a certain level of damage
occurring as a function of hazard intensity, with damage levels for build-
ings ranging from window failure to structural collapse and typically
categorised by construction type. A large range of variables influence
building response to volcanic hazards, including the age and condition
of the structure, the workmanship and construction quality and the
use of transient components such as shutters. The advantage of a vul-
nerability function, or curve, is that it accounts for some of the uncer-
tainty inherent in forecasting building response without a detailed
knowledge of each building and when faced with poorly understood
and complex damaging phenomena. Vulnerability functions are com-
mon in seismic risk assessments but less so in volcanic risk assessments
due to the infrequent and complicated nature of volcanic hazards and
the difficulties in reproducing the physical processes in the laboratory.
To date, physical vulnerability estimates have typically focussed on
one volcanic hazard (e.g. tephra fall: Spence et al., 2005) or, for the
few studies that are multi-hazard, one volcano (e.g. Vesuvius: Zuccaro
and De Gregorio, 2013). With this in mind, and drawing upon existing
knowledge and practice, our study focussed on providing standardised
survey methodologies and multi-hazard vulnerability estimates that
could then be updated and expanded following local field studies at
any volcano. This work comprised four main components, reflected in
the layout of this paper: firstly, an initial assessment of building vulner-
ability has been carried out for a set of broadly defined building types,
detailed in Section 2. Secondly, we considered that any loss or disrup-
tion to infrastructure (such as water, transport or electricity networks)
is likely to fundamentally impact local economies and livelihoods and
so infrastructure vulnerability should also be incorporated into any as-
sessment of potential eruption impact. Thus, in Section 3 we identify
critical infrastructure elements that support socio-economic activities
in volcanic areas and qualitatively discuss their vulnerability to the
key volcanic hazards. In the third part of the paper (Section 4) we dis-
cuss physical vulnerability survey methodologies and data require-
ments for more detailed study. In the final section (Section 5), we
apply our methodology to two case studies by refining the preliminary
vulnerability estimates of Section 2 through field surveys of the charac-
teristics and distribution of exposed buildings, the local agricultural
economy and infrastructure at Kanlaon volcano in the Philippines and
Fogo volcano in Cape Verde. These volcanoes were chosen from the
range of MIA-VITA (2009-2012)target volcanoes in developing regions
because there had been no previous assessments of the physical vulner-
ability of the surrounding areas to volcanic hazards and because they
represent differing volcanic settings and building stock profiles that
may offer insights into a wider range of analogous volcanoes.
2. Preliminary estimates of building vulnerability
Quantifying the vulnerability of buildings may be undertaken for
three primary reasons: firstly, to identify buildings that may benefit
most from mitigation measures that could be undertaken to ensure
the safety, as far as possible, of inhabitants (or livestock) who may be
trapped in buildings during a volcanic eruption; secondly, in cases
where the area has been successfully evacuated prior to an eruption,
to quantify the potential damage for loss estimation and rehabilitation
planning; and thirdly, to support the development of criteria or guide-
lines for construction of new buildings and modification of existing
buildings. The term vulnerability is used here in the engineering sense
as the likelihood of achieving a certain damage state given certain haz-
ard intensity, for example the likelihood of roof failure as a function of
ash fall load. Damage from volcanic hazards such as lava flows, lateral
blasts, sector collapse or debris avalanche damage can be considered
binary, with total damage in the areas impacted and zero damage in
areas not impacted, regardless of building type (Table 1). This may be
simplistic in some cases because, for example, lava flows can cause
fires outside of the zone of impact (Blong, 1984) and damage may be
gradational towards the peripheries of lateral blasts (Jenkins et al.,
2013).
Major volcanic hazards that do inflict gradational damage affect dif-
ferent components of a building (Table 1) and so require separate esti-
mates of vulnerability. Over the following subsections, we discuss the
damage and categorise major building types for each of the key volcanic
hazard parameters: impact energy (projectiles), horizontal load (ash
fall), lateral dynamic pressure (pyroclastic density current) and ground
shaking (volcanic earthquake). In many cases, building types borrow
from extensive seismic building stock data inventories (e.g. Zuccaro
and Papa, 2002; Spence et al., 2008; Jaiswal et al., 2011). In the ab-
sence of local building data, we provide preliminary estimates of vulner-
ability for each of the major building types and each of the hazards. To do
this, we considered the existing knowledge and literature, including em-
pirical data collected post-eruption, experimental data collected through
the testing of certain materials, theoretical calculations of building
strength and engineering expert judgement. This draws in particular
on several previous collaborative studies and surveys of building vulner-
ability to volcanic eruptions developed through the EU-funded European
Laboratory Volcanoes project (1994–1996: Casale et al., 1998), the
EXPLORIS Project (2004–2007: Marti et al., 2008; Spence et al., 2005;
Spence et al., 2007; Zuccaro et al., 2008), the SPeeD Project (2009–
2011: Jenkins and Spence, 2009a) and field observations following the
eruptions of Pinatubo in 1991 (Spence et al., 1996), Montserrat in 1997
(Baxter et al., 2005) and Merapi in 2010 (Jenkins et al., 2013). These vul-
nerability values act as a starting point for more refined estimates and
should be re-evaluated when new data become available, i.e. following
post-eruption impact assessment or engineering studies. Vulnerability
estimates based on the generic building types outlined in this section
should not be used within a risk assessment without a sufficient knowl-
edge and understanding of the local building stock characteristics and
distribution, typically identified through comprehensive field surveys,
to ensure the suitability of the classification.
2.1. Projectiles
Explosive eruptions produce hot projectiles that vary in size from
6.4 cm to 100 cm and can potentially impact more than 10 km from
the vent, but more typically land within 5 km (Blong, 1984). The impact
energy of projectiles is enough to puncture holes in roofs (Fig. 1), kill
people or livestock through blunt trauma, damage critical infrastructure
components and cause serious damage to crops. Projectiles not large or
dense enough to penetrate roofs can contribute to roof collapse through
overloading or through repeated impacts that may seriously weaken a
structure and leave it more vulnerable to future impacts or hazards. A
projectile’s impact energy (Ei, in Joules) and thus potential to cause
damage, is a function of its mass (m, in kg) and terminal velocity (v, in
m/s), as shown in Eq. (1). v in turn depends primarily on the height to
which projectiles have been ejected
1 2
Ei ¼ mv : ð1Þ
2
Blong (1981) and Pomonis et al. (1999) consolidate and summarise
empirical and experimental examples of blocks and bombs penetrating
roofs and identify relationships between roof materials and the impact
energy required to puncture such roofs for projectiles of varying size,
shape and density; we present salient results for six key roof classes in
Table 2, which can be considered preliminary vulnerability estimates.
Projectile impact to sheet or slab roofs is resisted by a single structure
over the whole roof area, while for tiled roofs each tile is a separate
structure with a relatively small area that may be damaged at lower im-
pact energies. The brittle nature of tiles also means that once they have
been damaged they will probably require replacement. Tiled roofs are
 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120 107

Table 1
For each of the primary damaging volcanic hazards, potential building damage, its primary cause and the relevant hazard parameter are outlined. This highlights the distinction between
describing damage parameters in a qualitative (column 3: Cause of damage) or quantitative (column 4: Hazard parameter) manner. V indicates damage types for which a quantitative
estimate of building vulnerability, i.e. a vulnerability function, is presented in this study for a range of likely building types.

Volcanic hazard Damage Cause of damage Hazard parameter V

Projectiles Roof and wall damage Impacts from projectiles Block size and velocity ✓
Projectiles or ash fall Roof damage Overloading from projectile or ash accumulation Load ✓
Fire damage Ignition of flammable structure components/ Temperature
building contents
Loss of use Burial Accumulated thickness
Ash fall Accelerated corrosion for metal components The acidic nature of ash pH
Blocked and/or damaged gutters and drains Ash accumulation Load and grain size
Damage to contents, especially mechanical Ingress of fine ash into buildings through Grain size
and electronic components ventilation or other openings
Damage to contents Ingress of mud and debris following rainfall Accumulated thickness
Pyroclastic density Building destruction Pressure impact Peak dynamic pressure with height ✓
current Wall and opening damage
Building destruction Puncturing from focussed pressure impact Missile size and impact energy
Wall and opening damage
Fire damage Ignition of flammable structure components Peak temperature and duration
or building contents
Loss of use Burial Deposit thickness
Lateral blast Wall damage Pressure impact Peak dynamic pressure, missile size
Building destruction and impact energy
Volcanic earthquake Damage to structure and contents Ground shaking Ground shaking intensity ✓
Foundation and vertical structure damage Ground deformation and/or landslip Ground deformation
Lahar Structural failure, wall damage or destruction Scouring of foundations, pressure impact from Dynamic and hydrostatic pressure,
flow and entrained missiles flow velocity
Damage to contents Failure or infiltration through openings Flooding and sediment deposition

Loss of use Burial Deposit thickness

Lava flow Total damage Burial, collapse or fire Thickness
Sector collapse, Total damage Burial and/or pressure impact Pressure and thickness
debris avalanche

therefore the most likely to experience damage or destruction following
impact from projectiles, assuming average to good condition for all
roofs. For specific areas or building types, detailed knowledge of
the characteristics influencing roof response, such as roof covering
minimum thickness or method of fixing, condition and the local build-
ing codes and practices can be used to improve estimates of their
vulnerability.
2.2. Ash fall
Roof collapse occurs when the static ash load exceeds the strength of
either the roof material, roof supporting structure or both (Fig. 2). Non-
structural elements such as gutters and overhangs are particularly
vulnerable to failure from ash loading and the load at which they fail
is dependent on roof area and drainage type as well as their material,
support and method of fixing. The load exerted by an ash fall depends
upon its density (ρ in kg/m3) and thickness (h in m) calculated as in
Eq. (2).
LAF ¼ ρgh ð2Þ
where LAF is the ash fall load in Pascals (Pa) and g the gravitational accel-
eration (9.81 m/s2). Factors to consider include recesses or sagging in
the roof that can exacerbate ash accumulation, and therefore load, in
weak spots and the load distribution across the roof, which may vary
through aeolian or manual removal, accumulation or compaction. The
pitch of the roof also affects a building's vulnerability to ash fall load; de-
pending upon the roof covering and condition, steep pitch roofs may
allow ash to slide off, analogous to snow falls. Ash clean-up actions
may worsen the damage caused by the static ash load, e.g. by people
falling off or through roofs. Precipitation can also increase the pressure
exerted by an ash load — by as much as 100% compared with dry ash
(Macedonio and Costa, 2012). Ash can also be corrosive to some

Fig. 1. Ballistic damage to metal sheet roof approximately 3 km from the vent of Pacaya volcano in Guatemala during the 2010 eruption. Maximum ballistic clast size in this area reached
25 cm.
Photo: Gustavo Chigna, INSIVUMEH.
108 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120
Table 2
Likely ranges of impact energies required by projectiles to penetrate each of the roof
classes, assuming they are of average condition (after Blong, 1981; Pomonis et al., 1999).
Penetration may still occur below the minimum value and some roofs may remain
intact above the maximum values but the ranges indicate where most of the damage
may be expected for typical roofing stock.
Roof class Thickness Range of impact energies
(mm) required for penetration
(Joules)
AP: Glass skylights/windows 2.5–4 0.15–2
BP: Tile (clay or terracotta) 10–40 10–80
CP: Fibre reinforced concrete sheet 4.5–9.5 20–85
DP: Timber boards 4.5–12 60–500
EP: Metal sheet 0.4–0.7 400–1000
FP: Reinforced concrete slab 50–125 4000–12,000
materials, particularly metal roofs, guttering and fittings, hence a long-
duration load can cause damage additional to that experienced in the
impact phase and, if the deposited ash is hot, it may weaken building
components, such as joists and attachments, or cause fires.
Damage assessments following the eruptions of Pinatubo,
Philippines in 1991 (Spence et al., 1996) and Vulcan and Tavurvar,
Papua New Guinea in 1994 (Blong, 2003) represent the only compre-
hensive studies of buildings affected by ash fall. Estimating vulnerability
from so few data points is problematic because as well as changes in
loading characteristics, strength for one roof class varies depending
upon the method of fixing between the covering and supports, their di-
mensions and local building codes and practices. Vulnerability estimates
have thus been supplemented with experimental and theoretical stud-
ies, with a strong focus on European volcanoes. Previous empirical, ex-
perimental and theoretical studies of roof failure under ash fall loading
are summarised in Jenkins and Spence (2009b) and based upon these
existing data and structural engineering judgement, we estimate ap-
proximate collapse load ranges for five major roof classes defined as
part of this study (Table 3). Roof collapse is considered to include failure
of the roof covering and/or of the supporting structures, e.g. purlins or
trusses. The roof typology, as well as roof spans and local practices,
will dictate which method of failure is more likely. An ash load that
does not cause roof failure can affect a building's response to subsequent
volcanic hazards, e.g. pyroclastic density current, because of the in-
creased lateral strength afforded to the vertical structural elements;
however, conversely this may cause failure from seismic shaking at a
lower shaking intensity although to what extent is not well quantified
(see Zuccaro et al., 2008 for potential vulnerability modifiers).
Larger agricultural, industrial and commercial buildings are typically
one storey and characterised by metal sheet roofs resting on purlins,
with clear distances of at least 5 m between primary roof supports or
Fig. 2. Roof damage from the 15 June 1991 eruption of Pinatubo volcano in the Philippines: a) Collapsed long-span metal sheet roof at Clark Air Base, approximately 20 km northeast of the
volcano; and b) Intact short-span metal sheet roof in the same area. Approximately 5 to 10 cm of wet ash accumulated in these areas during the 15 June 1991 eruption, equivalent to ap-
proximately 0.7 to 1.9 kPa load given measured ash densities (Spence et al., 1996).
Photos: USGS.
trusses. Such large-span buildings are generally more vulnerable to fail-
ure under ash loading than short-span structures (Fig. 2). In one area
surveyed during the Pinatubo 1991 eruption 75% of long-span roofs col-
lapsed compared to only 16% of domestic short-span metal sheet roofs
under approximately 0.7 to 1.9 kPa ash load (Spence et al., 1996). To ac-
count for these buildings, we defined roof class BAF (Table 3). Based on
the limited data available and following evidence from the Pinatubo
eruption, which suggest an approximately 30% increase in vulnerability,
we propose applying a vulnerability modifier of − 0.8 kPa relative to
roof class CAF: metal sheet roofs in average condition.
2.3. Pyroclastic density currents
Pyroclastic density currents (PDCs) are fast-moving currents of hot
volcanic ash and gases that typically comprise two main categories:
1) A block-rich concentrated flow that is largely valley-confined and
causes damage through burial, fire and/or lateral dynamic pressures
and entrained missiles.
2) A low- or high-energy dilute turbulent surge that is less confined by
topography and can move independently of the concentrated parent
flow.
Low-energy surges detach from a parent flow and largely cause dam-
age through thermal impact and ensuing fires. By contrast, high-energy
surges are produced by an explosive source process and characterised
by significant mechanical impacts from elevated dynamic pressures
that can fell trees and destroy structures. Damage to buildings and
other structures therefore depends upon the velocity, temperature, du-
ration and amount of solid material a flow carries, with greater damage
expected with the increase of each dynamic variable. In the absence of
fires, the major damage mechanism to buildings is likely to be lateral
dynamic pressures. Dynamic pressure and damage varies with the
velocity (v) and density (ρ) and its variation across a flow axis as well
as with the building dimensions, shape, height and orientation with
respect to the flow and the building's ability to shelter from the flow
behind other structures or topography. Generally, higher building vul-
nerabilities occur with increasing wall or panel span, height and de-
creasing thickness.
PDC dynamic pressures (P) can be calculated using Eq. (3), although
this ignores localised higher impact pressures that may be produced by
large objects (missiles) present within the current, such as boulders,
building debris or uprooted trees. The reader is referred to Spence
et al., 2007 for a detailed overview of missile impact.
1 2
P¼ ρv : ð3Þ
2
 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120 109

Table 3 study of opening vulnerability to lateral dynamic pressures has been un-
Roof classes defined as part of this study, their general characteristics and the estimated dertaken for building stock around Vesuvius, Italy (Zuccaro, 2000) and
mean and interdecile (central 80th percentile) range of ash fall loading required for roof
failure, assuming a Lognormal distribution with a standard deviation of 0.3 (see
involved theoretical and experimental studies. Additional estimates
Section 2.5). Roof failure may still occur below the minimum range value and some arise from the results of nuclear blast loading tests on buildings
roofs may remain intact above the maximum range values but the ranges indicate (Glasstone and Dolan, 1977) and from expert judgement (Valentine,
where most of the damage may be expected for typical roofing stock. 1998). Available results and our proposed failure ranges are summarised
Roof Description Collapse load for key opening types in Table 4. For glass, proposed failure pressures are
class (kPa) lower than those suggested by Glasstone and Dolan (1977) because the
Interdecile Mean response of brittle building components to nuclear blast loading is ex-
range pected to be different from that under PDC loading (Spence et al.,
AAF: Weak timber boards on timber rafters/trusses; 1.3 to 2.6 1.8
2004). For wooden frames, shutters and doors the relatively large sug-
Metal sheet roofs on timber rafters/trusses, in gested failure ranges reflect the increased uncertainty in considering a
old or poor condition; typology with generalised characteristics. The results in Table 4 suggest
Tiles on timber rafters/trusses, of old or that something as simple as boarding up openings prior to evacuation
poor condition;
may prevent internal damage or ignition of a building subject to pyro-
Vaulted masonry
BAF: Long span roofs with metal sheet or fibre 1.4 to 3.1 2.0 clastic density currents, as supported by observations of building damage
reinforced concrete sheets. on Montserrat (Baxter et al., 2005).
CAF: Metal sheet roofs on timber rafters/trusses, 1.5 to 5.3 2.8
in average condition;
Tiles on timber rafters/trusses, in average
2.3.2. Vulnerability of walls and panels
or good condition
DAF: Metal sheet roofs on timber rafters/trusses, 2.1 to 7.6 4.0 With increasing dynamic pressures, the overall building structure
in good condition; may fail, either through collapse of the frame, infill panels or, in the ab-
Strong timber on timber rafters/trusses in sence of a frame, the wall. As with ash fall, the building typology will dic-
average or good condition tate the method of failure. Pyroclastic density currents that impact
EAF: Flat reinforced concrete roof designed for 4.8 to 10.3 7.0
access and generally in good condition
densely populated areas are rare and much of the literature describing
wall or panel vulnerability to PDCs has been sourced from experimental
tests and theoretical calculations of wall and panel strength under lateral
dynamic pressures (e.g. Glasstone and Dolan, 1977; Valentine, 1998;
While lahars are not actively considered in this study because of the Zuccaro, 2000; Spence et al., 2001). For the most part existing vulnera-
limited empirical, experimental and theoretical data surrounding build- bility estimates are appropriate for European, and in particular Vesuvian,
ing response, an initial assessment of building vulnerability may be ob- building stock. However, empirical damage studies following the large
tained from pyroclastic density current vulnerabilities because in both explosive eruption of Merapi volcano, Indonesia, in 2010 (Jenkins
cases the principal damage mechanism is likely to be lateral pressures et al., 2013) offer some insight into vulnerabilities for weaker masonry
(both dynamic and hydrostatic). Further study is required to see if this buildings. The main limitation in generalising building typology is the ef-
is fully appropriate. In assessing building vulnerability to lateral dynamic fect of wall thickness and span, with thicker and/or shorter span walls
pressures, two building components are of particular importance: open- more resistant to lateral pressures. However, the age and condition of
ings (predominantly doors and windows) and walls or panels. a building and varying levels of construction and material quality, partic-
ularly for critical building components such as mortar, affect building
strength and therefore vulnerability. There are few quantitative data
2.3.1. Vulnerability of openings but timber and metal panels are likely more vulnerable than masonry
Openings are typically weaker than the surrounding built structure walls (Table 5). These classes can be common in volcanic areas because
and are thus likely to fail at lower lateral pressures; failure can occur material and construction costs are more affordable than masonry in
due to direct pressure from a PDC, from missiles carried in the flow or many volcanic areas and also because they are better suited to tropical
from thermal insult. The failure of one or more openings would allow a environments. Averaged failure pressure ranges in Table 5 have been
PDC to infiltrate into the building, which could trigger fires and would established from previous studies and expert judgement; more detail
likely be lethal for any trapped occupants. The most comprehensive is given in Jenkins and Spence (2009b).

Table 4
The load at which glazed windows, aluminium and wooden windows and doors fail, established from theoretical and experimental testing carried out by Laboratorio di Urbanistica e
Pianificazione Territoriale for building stock around Vesuvius: [1] Spence et al. (2004); [2] Zuccaro (2000); expert judgement: [3] Valentine (1998); and evidence from nuclear
explosions: [4] Glasstone and Dolan (1977). Suggested interdecile (central 80th percentile) ranges from this study are given in the final column, assuming a Lognormal distribution
(see Section 2.5) with standard deviation 0.5.

Opening types Dimensions Testing method Interdecile (mean) failure pressures

(kPa)

Literature Suggested

Glazing Glass of varied thickness and strength 0.5 m2 to 1.7 m2 Theoretical [1] 1.0 to 3.0 (1.8) Window:
Ordinary window glass Small and large Expert judgement [3] 1.0 to 4.0 (2.0) 1.0 to 4.0 (2.0)
Nuclear explosions [4] 3.4 to 6.9 (−)
Window frames, Aluminium window frame in bad condition 1.30 m × 1.60 m Experimental [2] 1.5 Aluminium window frame:
shutters Aluminium window frame in good condition 1.30 m × 1.60 m 3.0 1.0 to 4.0 (2.0)
Old wooden window shutters 1.00 m × 2.20 m 5.0 Wooden frames and shutters:
Window and door frames – Expert judgement [3] 7.0 to 17.0 (11.0) 4.5 to 17.0 (9.0)
Wooden window shutter – Nuclear explosions [4] 6.9 to 13.8 (−)
Wooden door Old wooden door 1.30 m × 2.65 m Experimental [2] 3.5 Wooden door:
Door – Expert judgement [3] 4.0 to 9.0 (6.0) 2.5 to 9.5 (5.0)
Wooden door – Nuclear explosions [4] 2.8 to 9.0 (−)
110 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120

Table 5
Wall and panel classes relevant to failure under PDC dynamic pressures, their general characteristics and their estimated mean and interdecile (central 80th percentile) range of failure
pressures, in kPa, assuming a Lognormal distribution with a standard deviation of 0.3 (see Section 2.5). The range indicates where most of the damage may be expected and so failure
may still occur below the minimum range value and some walls or panels may remain intact above the maximum range values.

Wall, panel Description Failure pressure Data source

class (kPa)

Interdecile Mean
range

APDC: Metal sheet panel in a timber frame 1.5 to 4.0 2.5 This study
BPDC: Timber panel in a timber frame 2.0 to 4.5 3.0 Spence et al. (2004); This study
CPDC: Unreinforced masonry, brick, block or stone; 3.5 to 8.0 5.5 Jenkins et al. (2013); Spence et al. (2004);
Reinforced concrete wall or panel without earthquake resistant design Zuccaro (2000)
DPDC: Reinforced masonry, concrete panel or confined masonry with 6.5 to 15.0 10.0 Glasstone and Dolan (1977); Pomonis et al. (1999);
earthquake resistant design; Spence et al. (2004)
Thick (N0.5 m) rubble stone masonry

2.4. Volcanic earthquakes
Volcanogenic earthquakes occur prior to an eruption due to the up-
ward movement of magma, but also during eruptions due to shallow
gas explosions, ground fractures and caldera or dome collapse (Schick,
1981). Their shallow focus (typically b 5 km) results in strong and dam-
aging ground shaking within a few kilometres of the vent but rapid
attenuation with increasing distance. The most damaging impact of
earthquake shaking for buildings is failure of masonry walls or the struc-
tural frame. Racking of the frame, where shaking moves floor levels lat-
erally with respect to each other, may cause wall and roof collapse
(particularly for older masonry buildings), severed connections be-
tween the roof, walls and floor or large cracks in the walls, frame col-
umns or beams. Rupturing of the gas lines, electrical circuits and/or
water pipes may cause subsequent fire and flooding, which can result
in more damage than the initial shaking. Few definitive examples of vol-
canic earthquakes causing damage exist because distinguishing histori-
cal accounts of tectonic and volcanic earthquake damage is difficult.
Moment magnitudes of 5 or below are most likely and so less damage
may be anticipated than for tectonic earthquakes (Zobin, 2001), how-
ever the damage potential is not negligible and we discuss it here for
completeness.
Many studies have been made regarding the vulnerability of build-
ings to tectonic earthquakes. Following Grünthal (1998), we have
assigned five building classes and consider their probability of partial
or total structural failure (≥ D4: Grünthal, 1998) as a function of EMS-
98 felt intensities (Table 6). Likely intensities range from V (strong
vibration, buildings tremble throughout) to VIII (heavily damaging). Es-
timates borrow from the Global Earthquake Vulnerability Estimation
System (GEVES: Spence et al., 2008), which considers a number of
global and regional studies as well as unpublished data at Cambridge
Architectural Research, in concert with the data quality, to evaluate seis-
mic vulnerability for a set of building classes. Given differences in build-
ing characteristics, there will be significant variability within each class,
e.g. reinforced concrete frame buildings may or may not be constructed
with earthquake resistant design. Further refinement towards a specific
building stock is therefore appropriate; however, timber and steel
frame buildings are likely to remain relatively less vulnerable than ma-
sonry or reinforced concrete frame buildings due to their flexible and
ductile nature.
presented, using a distribution that is Lognormal in form due to the mul-
tiple sources of uncertainty (Eq. (4)).
  
P failurejIÞ ¼ Φ ln I; ln I f ; σ ð4Þ
where Φ in Eq. (4) is the standard form of the cumulative Normal distri-
bution, which we define in terms of the natural logs of the given hazard
intensity (ln I) and the mean intensity required for failure (lnIf). A stan-
dard deviation, σ, that represents a best average fit to available data
should be used and, based on our studies of existing data variability
(Jenkins and Spence, 2009b), as an initial estimate we suggest a value
of at least 0.3, or 30% of the mean value. For simplicity this should be
kept uniform across building types, unless there are data to suggest oth-
erwise. This form of vulnerability curve is widely used in vulnerability
studies in earthquake engineering (e.g. Yamazaki and Muroa, 2000;
Rossetto et al., 2013). Such vulnerability and uncertainty values should
be re-evaluated when considering the vulnerability of specific building
stock (see Sections 4 and 5).
3. Vulnerability of critical infrastructure
Much less data are available regarding the vulnerability of infra-
structure to volcanic hazards than for buildings. However, in many
parts of the world the failure, disruption or reduced functionality of
Table 6
Building structure classes with reference to potential earthquake shaking damage, their
general characteristics and probability of partial or total structural failure with
increasing felt shaking intensity (EMS-98). Vulnerability estimates borrow from
unpublished data used in the GEVES study for building vulnerability to tectonic earth-
quake (Spence et al., 2008).
Building class Description
AEQ: Weak rubble stone or fieldstone masonry with no frame;
adobe (earth brick)
BEQ: Load-bearing simple stone or manufactured stone masonry
with no frame
CEQ: Reinforced concrete frame or walls; confined masonry
DEQ: Steel frame
EEQ: Timber frame

2.5. Developing vulnerability functions Building class Mean probability (%) of partial or total structural
failure at EMS-98 intensity:
Estimates of vulnerability must be considered with appropriate error V VI VII VIII
bars, which may be reduced with comprehensive building surveys, new AEQ 0.6 4 17 40
data and a reconsideration of available data. Based on available data and BEQ 0.0 0.4 4 15
the authors' judgement, we propose that preliminary vulnerability CEQ 0.0 0.3 2 7
functions describing the probability of failure, P(failure), as a function DEQ 0.0 0.3 2 4
EEQ 0.0 0.3 1 4
of hazard intensity (I) could be derived from the mean failure values
 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120 111

Table 7
Potential impacts from ash fall for critical infrastructure. Minimum intensity thresholds for impacts have been delineated where possible, following Blong (1984) and Wilson et al. (2012).
Damage at higher intensities (N100 mm) includes those experienced at lower intensities (N1 mm). For example, at ash thicknesses of 100 mm there is the potential for failure of road
drainage systems as well as for vehicle accidents due to poor visibility and reduced traction.

Infrastructure component Potential impacts from ash thickness exceeding:

1 mm 10 mm 100 mm

Machinery Vehicles ▪ Paintwork corrosion and abrasion ▪ Increased wear of engine and brakes
▪ Clogging of air filtration systems
▪ Windscreen abrasion
Electrical and ▪ Abrasion and clogging of some ▪ Damage to external building electronics
mechanical air-handling, mechanical and (e.g. security systems)
components electrical equipment
▪ Corrosion of metallic elements
Transport Roads ▪ Poor visibility ▪ Reduced traction ▪ Potential failure of drainage systems,
▪ Road markings obscured ▪ Clogging of roadside drains and ditches embankments and bridges
▪ Reduced traction
Rail ▪ Abrasion of engines ▪ Short-circuiting of overhead power lines ▪ Track drainage problems
▪ Corrosion of metallic parts ▪ Power failure ▪ Failure of station buildings through
▪ Shutdown of system through ▪ Failure of signalling equipment ash overload
operational problems
Power Generation ▪ Short-circuiting of power lines ▪ Failure of generating plant (dependent on
(particularly with wet ash) system type and design)
Distribution ▪ Flash-over of power lines ▪ Abrasion, clogging and flash-over causing ▪ Structural damage to distribution
(particularly with wet ash) disruption and/or damage to some electrical lines and support structures
▪ Flashover of transformer insulators and mechanical equipment at substations ▪ Failure of tree branches due to ash
(particularly with wet ash) loading affecting distribution lines
Telecommunications Radio ▪ Communication interference by
volcanic cloud lightening
▪ Possible signal attenuation
Mobile phone ▪ Overload of system through ▪ Damage to external telecommunication ▪ Damage to communication dishes
networks high demand components and power cables through and microwave towers due to excess
flash-over, abrasion and/or corrosion ash loading
Landline ▪ Un-insulated lines may flash-over ▪ Exchange air-cooling systems disrupted ▪ Damage to overhead lines and
networks ▪ Overload of system through or damaged support structures
high demand. ▪ Damage to lines from tree branch
failure due to excess ash loading
Water systems Supply ▪ Increase in turbidity and acidity of ▪ Possible chemical contamination of
(drinking and surface water storage surface water storage
irrigation) ▪ Exhaustion and restrictions of ▪ Clogging of uncovered intakes, channels
water supply due to damage and/or and treatment plants
increased demand ▪ Damage to pumping equipment
Waste ▪ Clogging of storm drains and sewers ▪ Equipment failures in sewage treatment (where
▪ Damage to waste water treatment not separated from storm water drainage)
plants, such as pumping equipment, ▪ Ash overload in treatment plants
hydraulic seals, etc.

infrastructure is likely to have a larger impact on livelihoods and (interdependencies between physical, economic and social systems).
the local economy than damage to buildings. In some cases it can We therefore recommend that future physical vulnerability assess-
act as a catalyst to existing economic, social or agronomic decline ments for volcanic areas should always include surveys of exposed
(e.g. Wilson et al., 2011) because of a high systemic vulnerability infrastructure.

Table 8
Potential impacts from pyroclastic density currents and lahars for critical infrastructure.

Potential impacts from pyroclastic density currents and lahars

Machinery Vehicles ▪ Vehicles buried, destroyed or swept away

Electrical and mechanical components ▪ Damage, destruction by high temperatures and/or pressures and burial of external surfaces
Transport Roads ▪ Road surfaces damaged by heat
▪ Road signage and electrical signalling destroyed
▪ Bridges destroyed, riverbanks eroded and fields flooded where riverbanks overtop
▪ Roads, channels and bridges buried
Rail ▪ Engines, cars, tracks and stations damaged and destroyed
▪ Wooden sleepers burnt; track buckled, buried or swept away
▪ Overhead power lines destroyed
Power Generation ▪ Damage to hydroelectric dams by over-topping or significant sedimentation of the reservoir
▪ Damage to generation building and equipment
Distribution ▪ Destruction of transmission and distribution power lines and substations
Telecommunications Radio ▪ Destruction of electronics and structures if station struck
Mobile phone networks ▪ Damage unlikely, except in the case of pyroclastic surge, as antennas are based on topographic highs
Landline networks ▪ Destruction and burial of ground level and overhead cabling
Water systems Supply ▪ Thermal destruction of plastic pipe work; damage to metal pipe work where not underground
▪ Infilling of stream channels cause overbank flows where riverbanks overtop
▪ Destruction of plastic pipe work; damage to metal pipe work where not underground
▪ Sedimentation and contamination of uncovered catchments
Waste ▪ Blockage of drains and damage to waste water treatment plants by deposits
112 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120

Table 9
Potential impacts from earthquake shaking for critical infrastructure.

Potential impacts from earthquakes N EMS-98 felt intensity V

Machinery Vehicles ▪ Secondary damage caused by structural failure, e.g. buildings collapsing onto vehicles
Electrical and mechanical ▪ Secondary damage caused by structural failure
components
Transport Roads ▪ Structural damage to some masonry bridges, slippage of embankments and bridge abutments.
▪ Structural cracks from ground deformation
▪ Roads can become blocked from collapsed structures
Rail ▪ Minor damage to track and overhead lines
Electricity systems Distribution ▪ Minor damage to power lines
Telecommunication systems Mobile phone networks ▪ Structural failure of antennae
Landline networks ▪ Secondary damage caused by structural failure
Water systems Supply ▪ Possible damage to reservoirs, dams and intake structures, allowing water leaks
Waste ▪ Sewage and storm water pipes may be ruptured by shaking or ground deformation, contaminating local areas

The impacts of a volcanic eruption for infrastructure vary depending
upon the volcanic hazard and its intensity but could include disruption
of electricity supplies, contamination of agricultural processing areas
and sedimentation of surface water networks, requiring extensive and
repeated clean-up. Cascading effects such as delays in transport times
are then likely to compound any disruption and associated impacts.
Transport loss of function due to volcanic products being deposited on
the route can potentially be mitigated through the use of engineered
channels, dams and barriers (PDC and lahar) or repeated clean-up
(ash). Such impacts are likely to occur with lower intensity hazards
than damage to buildings. However, the diverse range of infrastructure
system design, types and configurations make it very difficult, perhaps
impossible, to reliably create generic infrastructure vulnerability curves.
Therefore analysing interdependencies between infrastructural systems
and carrying out comprehensive local inventory surveys to produce
site-specific vulnerability functions is the most valid approach. Here,
we provide a qualitative overview of infrastructure vulnerability to
each of the major volcanic hazards and, in Section 5.5, briefly discuss in-
frastructural vulnerability for the two case study areas.
3.1. Ash fall
Volcanic ash consists of tiny fragments of rock and volcanic glass less
than 2 mm in diameter. Thus ash can infiltrate many openings, clog air-
filtration systems and abrade or scratch most surfaces, especially be-
tween moving parts of vehicles. Ash falls are unlikely to cause structural
damage to transport networks, except in the case of weak bridges, how-
ever they can cause significant disruption to the functioning of transport
routes through reduced visibility and traction, at thicknesses as small as
1 mm (e.g. 2011 Shinmoedake eruption, Japan: Magill et al., 2013). The
abrasive, corrosive and conductive (especially when wet) properties of
ash can also damage non-structural infrastructure components under
similar thicknesses (Blong, 1984; Blong, 2000; Wilson et al., 2012),
with the level of damage dependent upon the physical and chemical
properties of the ash. Due to the high vulnerabilities of infrastructure
under relatively small ash fall, impacts to water, electricity and
telecommunication systems are more likely to result from ash fall
than from any other volcanic hazard. Table 7 summarises ash fall im-
pacts for the five key infrastructure components that support activities
in both rural and urban areas. For a more detailed review of ash fall im-
pacts the reader is referred to Wilson et al. (2012).
3.2. Pyroclastic density currents and lahar
The significant mechanical and thermal impact of pyroclastic density
currents (PDCs) or lahars means that most infrastructure in their path
will be destroyed. Bridges are especially susceptible to damage as
gravity-based flows follow topographic depressions, i.e. areas requiring
bridge crossings. Identifying lahar or PDC channels and strengthening
exposed bridges can reduce bridge vulnerability in future events but
for affected transport routes, they may be impassable until deposits
are cleared away: a considerable time in the case of hot pyroclastic or
lahar deposits. Ash fall, PDC or lahar deposits not cleared from roadsides
may clog ditches and culverts, which during rainfall can then cause ero-
sion of the shoulder and road surface. Other potential impacts from PDC
or lahar are summarised in Table 8.
3.3. Volcanic earthquakes
Volcanic earthquakes typically cause structural failure or collapse,
with embankments, masonry bridges and bridge abutments vulnerable
to slippage. Structures that have been built to seismic codes are signifi-
cantly less vulnerable than those without seismic construction codes
but fire following earthquake due to ruptured gas and power lines can
still cause total damage to structures not destroyed by the original shak-
ing. Table 9 outlines potential impacts of earthquake shaking for
infrastructure.
4. Field survey methodologies
The preliminary quantitative (buildings) and qualitative (infrastruc-
ture) estimates of physical vulnerability to volcanic hazards described

Table 10
Key survey methodologies used in assessing characteristics that may affect a building's physical vulnerability to volcanic hazards.

Survey type Description

Comprehensive Inventory surveying of every building, or every building of a certain use (e.g. residential), at risk in a population settlement. This is most useful for
communities that are at very high risk or for which the consequences of damage are significant. Such surveys are generally very time-consuming.
Single-point sampling Surveying a fixed number of the closest buildings to a point within a settlement (commonly 20, depending upon the size of settlement). These surveys
are commonly conducted when resources are limited. They give an indication of the building typology distribution within a settlement where blanket
surveys are not feasible.
Multi-point sampling Surveying a fixed number of the closest buildings to multiple points within a settlement. For larger settlements, a single point survey may not be sufficient,
particularly where there are distinct areas that are likely to be characterised by different typologies, for example a historical centre, a business district.
Special interest Additional buildings may be individually surveyed to include specific sites of interest, e.g. potential evacuation shelters, emergency management centres,
churches, hospitals, schools, historical buildings or industrial areas for which the consequences of damage for local communities may be very large.
Transect Surveys that focus on lines that extend radially from the vent should take into account the expected hazard footprints outlined in the hazard map,
for example focussing on forecast ash deposition or pyroclastic density current overspill.
 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120 113

Table 11
Field data requirements and rationale for assessing building vulnerability to volcanic hazards. Many requirements will be refined for a specific area and over the course of the survey and
these overview requirements build on two decades of surveying experience at Cambridge Architectural Research and colleagues (Pomonis et al., 1999; Zuccaro et al., 2008).

Data to be collected Rationale

Location, GPS and area: Collected for reference

Building ID assigned:
Number of photos:
Date:
Usage: Building typologies may sometimes be categorised by use, for example schools may be built in a similar fashion. Identifying the use also
allows the assessor to consider the suitability of larger buildings as shelters.
Building age: May indicate the likely structural condition and also may allow for the identification of buildings constructed according to building codes
specifying different levels of (wind or earthquake) loading. Where public records aren't available, categories such as new, intermediate
and old, accounting for the range of local building ages, are best defined during the survey.
Building footprint: Useful in establishing an indication of reconstruction costs.
Number of storeys: Influences the damage costs associated with reconstruction, and the potential for casualties. This information may also be used to indicate
the resistance of a building to lateral dynamic pressures, from pyroclastic density currents, lahars or blasts.
Wall structure: The material and construction of building walls offer insights into their resistance to seismic action or the dynamic pressures emplaced by
pyroclastic density currents, blasts or lahars. Where possible the wall or panel height and length should be recorded as this will affect the
mechanism, and subsequent pressure, at which failure may be expected.
Roof structure: The main damaging action to buildings from ash falls is structural damage of the roof. The material used for the roof covering and supports,
as well as their shape and pitch and the construction methods and quality, e.g. spacing of supports and roof ties, inform their vulnerability.
Orientation with respect to volcano: For a geolocated building, the orientation can be important with respect to volcanic flow impact, evacuation direction and may also aid
identification of the building on satellite images.
Number, type and size of openings: Openings represent the most vulnerable elements of a building structure to dynamic pressures and are the first to fail.

above were developed to provide a foundation for local vulnerability as-
sessments to build upon. Depending on time, resources and funding
there are a number of different survey methodologies to identify local
building and infrastructure characteristics with respect to their vulner-
ability to volcanic hazards (Table 10). Field surveys discussed here, and
used in our case study areas, build upon previously described survey
methodologies (e.g. Marti et al., 2008; Spence et al., 1996; Zuccaro
et al., 2008). We recommend collecting data about the primary charac-
teristics of buildings, such as construction material and method, as well
as subsidiary characteristics that may be important in assessing impacts,
for example the size and prevalence of openings (Table 11). We do not
provide structured field survey methods for infrastructure but recording
the location, characteristics and number of each vulnerable item as
discussed in Section 3 is advisable. Survey locations should be chosen
as a function of the forecast hazard footprints and the community and
environmental setting, e.g. urban, rural, proximal or distal to the volca-
no. As a minimum, the field data outlined in Table 11 should be collect-
ed. It is expected that given a standard survey group of six people and
field trip duration of seven days, approximately 20 to 25 building
surveys of 20 buildings each can be conducted. Depending on the num-
ber of variants in the population of buildings in the survey area, this may
be expected to give a reasonably representative sample. Given no re-
strictions to resources, as many building types as possible should be
sampled, representing varying construction standards, ages and condi-
tions, as well as different geographical, cultural and socio-economic
areas.

Fig. 3. Location of Negros Island in the Philippines, with inset: volcanic hazard map for Kanlaon volcano (PHIVOLCS, 2003) overlain with the ten building survey locations.
114 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120
Fig. 4. Location of Fogo Island in the Cape Verde archipelago, with inset: provisional volcanic hazard map for a large explosive eruption similar to the 1680 event (Day and Faria, 2009)
overlain with the eleven building survey locations.
In addition to the components described in Table 11, it can be valu-
able for building surveys to collect extra data, where possible, such as
any building design codes or building components that can affect the
potential damage, e.g. air-conditioning units (potential failure), gas cyl-
inders (fire potential) and outdoor furniture or firewood (missile and
fire potential). Many of the survey categories may be best refined during
the course of the survey, and with a greater familiarity of the local build-
ing stock, however it is important to remain consistent over multiple
surveys. A short pilot survey to determine the final survey categories
may therefore be advisable. For each building, at least one geolocated
photo should be taken so that post-survey analysis is possible and so
that there is a record of the building in its state at the time of surveying.
Building types can be approximately identified through aerial photos,
and later verified by ground surveys, allowing large areas to be remotely
surveyed with limited costs and time. In the absence of aerial photos,
Google Earth images, or even in some areas Google Street View or street
maps may be useful. Electronic mapping of vulnerability in a GIS system
is recommended, either by using a digital survey form in the field or by
matching survey data with the geolocated building. Alongside maps of
the hazard and exposed elements, this can be valuable in visualising
the overall risk and the contributors therein, lending itself to GIS-
based risk modelling (e.g. Biass et al., 2013).
5. Case studies
We used the above field survey methodologies (Section 4) and
vulnerability estimates (Sections 2 and 3) to create building and
infrastructure inventories (Table 11) that could be used in
assessing physical vulnerability at Kanlaon volcano in the
Philippines and Fogo volcano in Cape Verde. We focussed our stud-
ies on these volcanoes because local partners identified that no vul-
nerability studies had been carried out and little was known about
the potential volcanic risk in these areas and because each volcano
offered different physical, socio-economic and agricultural
settings.
Kanlaon volcano in the Philippines forms part of the Negros volcanic
arc in the Central Philippines (Fig. 3). Six major rivers run from the
slopes of Kanlaon to the west coast of Negros creating an efficient drain-
age system that carries fertile volcanic deposits and deposits them on
the lower slopes. This makes Negros Island very suitable to agriculture,
which supports much of the population through subsistence farming,
small-scale trade and large commercial ventures such as fruit farms,
rice and sugar mills. Future eruptions are expected to either continue
producing the small magnitude ash emissions recorded since 1866 or
return to the larger magnitude explosive or effusive eruptive behaviour
indicated by the geological history (Martinez-Villegas et al., 2001).
Based upon the current topography and distances reached by geological
pyroclastic flow units, PHIVOLCS have highlighted areas to the south
and west of the volcano – particularly the communities of Cabagnaan,
Biak-na-Bato and Masolog – as areas exposed to pyroclastic density cur-
rents (PDC) and lahar (Fig. 3). A significant number of dwellings around
Kanlaon, in some places along river valleys or within the permanent ex-
clusion zone, are built on public land without title. Such settlements are
illegal and their number and vulnerability very difficult to calculate.
 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120 115

Fig. 5. The four predominant building types identified at Kanlaon volcano, Philippines, through field survey; these cover 96% of the surveyed buildings: a) Nipa dwellings: timber frame
with bamboo weave or timber infill, one or two storey basic rural dwellings with nipa (palm frond) roofs, b) Timber frame dwellings: timber frame with bamboo weave, timber or metal
infill, one or two storey buildings with corrugated steel roofs, c) Reinforced concrete frame buildings: generally one to two storey, but possibly up to four in more built up areas, with ma-
sonry infill walls and corrugated steel roofs, d) Mixed construction buildings: One to two storey buildings generally comprising a reinforced concrete lower half and a timber frame, timber
infill upper half with a corrugated steel roof. For buildings of mixed construction it is advisable to class them according to their most vulnerable construction type.

Fig. 6. The four predominant building types identified at Fogo volcano, Cape Verde, through field surveys; these cover 97% of the surveyed building types: a) Rubble stone masonry building
shown here with a reinforced concrete roof, although some had metal or tiled roofs; b) Confined masonry building with a reinforced concrete roof; c) Cut block masonry building with no
frame and shown with a reinforced concrete roof, although a small number had tiled roofs; d) Reinforced concrete frame building with a reinforced concrete roof. The majority of buildings
were one storey high, although in more built up areas buildings with four or more storeys were observed. Tiled roofs are sometimes constructed above reinforced concrete slabs to aid
runoff: the underlying slab best describes the expected vulnerability of such roofs.
116 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120

Table 12
Derived roof class vulnerabilities for penetration under projectile impact and collapse under ash fall loading, relevant to the surveyed building stock at Kanlaon and Fogo volcanoes. Small
proportions (2% at Kanlaon and 5% at Fogo) of the buildings surveyed were of unusual construction, e.g. a combination of the defined types; in these cases the more vulnerable classes here
should be used to represent such buildings. Intensity values represent failure of the roof: thus damage is likely at lower intensities than those proposed here.

Roof class Description Interdecile failure intensity range

(% of surveyed) (mean):

Kanlaon Fogo Projectiles Ash fall

(Joules) (kPa)

K_AP/AF (24) – Nipa thatch on timber supports; 70 to 150 (100) 1.0 to 2.1 (1.4)
Corrugated steel sheets in poor condition (N75% visible corrosion) on timber supports
K_BP/AF (34) F_AP/AF (8) Terracotta tiles on timber supports 70 to 150 (100) 1.4 to 3.1 (2.1)
– F_BP/AF (8) Corrugated steel sheets in average condition (between 25 and 75% visible corrosion) on timber supports 500 to 1100 (750)
K_CP/AF (41) F_CP/AF (6) Corrugated steel sheets in good condition (b25% visible corrosion) on timber supports 700 to 1500 (1000) 2.2 to 4.8 (3.3)
– F_DP/AF (74) Flat reinforced concrete slab roof with unknown amount of reinforcement and relatively low strength concrete 3000 to 6000 (4000) 3.4 to 7.3 (5.0)

They were not examined in this study: future surveys may look to inves-
tigate the vulnerability of such communities.
Fogo Island, part of a hotspot archipelago approximately 600 km
west of Africa, comprises a large stratovolcano with the summit peak,
Pico do Fogo, sitting within a 9 km wide lateral collapse structure
(Day et al., 1999) (Fig. 4). Major industries on Fogo Island include agri-
culture, fishing and, more recently, tourism. The 1995 flank eruption of
Pico de Fogo prompted the permanent evacuation of villages in Chã das
Caldeiras (Fig. 4) but these have since been repopulated and the
Chã floor is now home to approximately 800 people as well as the
burgeoning wine and tourism industry. Official hazard maps are not
publicly available for Fogo, however Day and Faria (2009) identify
three scenarios in an outline hazard analysis: 1) A continuation
of post-1785 effusive activity on volcanic fissures within Chã das
Caldeiras; 2) More intense and frequent fissure eruptions similar to
those during the 17th and 18th centuries; and 3) A large explosive
eruption from the summit of Pico do Fogo as in 1680. In addition, the
neighbouring island of Brava, 20 km to the west of Fogo shows geolog-
ical evidence of Plinian volcanism and in certain wind conditions large
eruptions from Brava may be expected to deposit ash falls on Fogo
Island. A large explosive eruption from Fogo was considered most perti-
nent to this study.
Field surveys in both areas were tasked with identifying building
characteristics and other at-risk elements that may affect a community's
capacity to recover from an explosive volcanic eruption. Over the next
sections, we present our findings and refine the broader vulnerability
assessments of Section 2. These vulnerability assessments formed part
of a larger effort to improve risk assessment at Kanlaon and Fogo,
including installing monitoring equipment (e.g. Faria and Fonseca,
2014) and improving hazard assessment (e.g. Neri et al., 2013).
5.1. Building characteristics
We carried out single-point surveys (Table 10) of 325 buildings in 21
communities around Kanlaon (10 locations) and Fogo (11 locations)
volcanoes, which we then used to estimate the distribution of building
typologies across the exposed areas. Survey sites were chosen on the
basis of their exposure to likely volcanic hazards (Figs. 3 and 4) and as
a function of the fieldwork resources and time available. These surveys
allowed us to derive vulnerability estimates for applicable local building
typologies for each of the key volcanic hazards. Vulnerabilities build
upon the initial estimates outlined in Section 2 by incorporating addi-
tional data relevant to the categorised building types. For more details
the reader is referred to the MIA-VITA technical report of Jenkins and
Spence (2010).
Four predominant building types have been identified at each of
the volcanoes (Figs. 5 and 6). There is no clear pattern to the spatial dis-
tribution of typologies at either volcano, however timber frame and
panel buildings with steel sheet roofs dominate at Kanlaon (56% of
buildings surveyed) and masonry buildings with flat reinforced con-
crete slab roofs at Fogo (75% of buildings surveyed). A more comprehen-
sive survey covering the range of built morphologies, e.g. urban,
suburban and rural, might have highlighted relationships for the distri-
bution of building types and conditions across different economic, geo-
graphic or cultural settings but this was beyond the scope of our studies.

Fig. 7. Breakdown of the opening types identified through field surveys in the Kanlaon and Fogo areas, ordered from most (left) to least (right) vulnerable to lateral pressures.
 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120 117

5.2. Vulnerability to projectiles and ash fall
Steel sheet roofs represent a traditional and economical construction
material in many volcanic areas, including at Kanlaon and Fogo, and
when newly constructed they are relatively strong. However, a tropical,
humid and coastal environment results in much faster corrosion of the
roofs and supporting metal purlins than in temperate climates and
thus they become brittle and their vulnerability increases as they age
(Table 9). Steel sheets that have corroded have a much higher friction
coefficient than good condition steel sheets, as do thatch roofs, and
both may be expected to retain ash loads more easily than new steel
sheets, increasing their likelihood of failure. Roof supports in residential
buildings are typically designed to at least withstand the load associated
with someone working on the roof. The load associated with someone
carrying out roof repairs or adding roof covering may be approximately
1 kPa (100 kg over 1 m2) and thus failure loads somewhat larger than
that, at 1 to 2 kPa, are likely for the weaker roofs (Table 12). Failure of
reinforced concrete slabs is more likely to occur from flexural yielding
of the structure surrounding the point of impact than from punching
shear associated with penetration of the structure. Structural failure of
RC slabs is dependent upon the amount of reinforcement, thickness
and quality of concrete, which can usually only be determined at the
time of construction or with destructive testing. RC slab roofs around
Fogo showed evidence of air gaps within the concrete and were rela-
tively thin (~15 cm), allowing limited room for reinforcement. This is
therefore reflected in our vulnerability estimates (Table 12).
5.3. Vulnerability to pyroclastic density currents
5.3.1. Openings
Differences in climate between the two study sites determined
which opening typologies are more likely (Fig. 7). For example, many
openings in the Kanlaon area have no covering (30%), most likely a fac-
tor of the tropical climate, while openings in Fogo, which experiences
colder night-time temperatures than Kanlaon, were mostly timber
board (45%). Timber covered openings, predominantly in the form of
shutters, show one of the lowest vulnerabilities to dynamic pressures
(Table 4), with experimental testing carried out in the Vesuvius area
showing mean collapse loads of around 5 kPa (Zuccaro, 2000). How-
ever, opening resistance depends strongly on the effectiveness of the
connection to and strength of the adjacent wall, which is likely to be sig-
nificantly poorer in these case study areas compared with the Vesuvian
area. For buildings at Kanlaon and Fogo where shutters were not used,
openings were typically single-glazed glass, which is significantly
more vulnerable to failure under lateral pressures. Depending upon
the size of opening, thickness of glass and condition of window frames,
PDCs with lateral dynamic pressures lower than 1 kPa (Table 4) may
cause the failure of such openings.
5.3.2. Walls and panels
In the Kanlaon area, timber panels included both thatch-covered and
weak plywood/hardwood boards, which are likely to offer a low resis-
tance to lateral dynamic pressures (Table 13) as well as being highly
flammable. For economic reasons timber panels seem to dominate in
rural areas, while in more urban areas less vulnerable masonry-infilled
reinforced concrete frame (RCF) are prevalent. RCF buildings, common
at both Kanlaon and Fogo, are considerably more expensive to build
than unreinforced block or rubble stone masonry buildings and their
distribution around the volcanoes is varied. Unfortunately, during a
building survey it is often difficult to differentiate buildings constructed
to a seismic code, where an attachment is included between the infill
masonry and reinforced concrete frame providing a better response to
lateral pressures, from those not constructed to a seismic code; and so
building construction records are critical. Following discussions with
government engineers at the case study sites, some RCF buildings are
expected to follow standards for earthquake resistant design; however,
even where the standards have been adequately implemented and
enforced, different construction standards and codes in the Philippines
and Cape Verde means that we may expect buildings with or without
earthquake resistant design to show higher vulnerabilities than analo-
gous European construction and a pressure modifier has been used to
estimate appropriate failure pressures (Table 10: Jenkins and Spence,
2010). In the absence of definitive knowledge about whether certain
RCF buildings are constructed to earthquake resistant standards, non-
earthquake resistant standards should be assumed. Rubble stone ma-
sonry, which includes nearly a third of the buildings surveyed at Fogo,
are typically made from strong locally-sourced scoria and offer signifi-
cant resistance to lateral pressures because of the stone's thickness
(~40 cm) and weight. However, surveys at Fogo found that the blocks
do not appear to have any mortar infill, which in turn reduces their re-
sistance to lateral pressures (Table 13) relative to experimental pressure
testing of analogous 40 cm volcanic masonry walls in Italy (mean failure
pressure of 7.9 kPa: Zuccaro, 2000) and preliminary estimates (Table 5).
5.4. Vulnerability to volcanic earthquake
A relatively strong maximum felt intensity (EMS-98) of 8 may result,
albeit with low probability, from volcanogenic earthquakes at Fogo and
Kanlaon. Building classifications with respect to earthquake shaking
(Table 14) at Kanlaon and Fogo are similar to those defined for PDC im-
pact, although their relative vulnerability differs. The majority (62%) of
buildings at Kanlaon have timber frames built using light timber (in
some cases, bamboo), which tend to be flexible enough to withstand
high intensities although connections between the vertical frame and
roof supports can be weak. It is likely that wind loads associated with
typhoons will impart a greater force upon these buildings than from
volcanogenic earthquakes. Older timber framed buildings around
Kanlaon, where the nipa thatch and timber have become more brittle,
are expected to be significantly more vulnerable to earthquake due to
their reduced flexibility. We therefore assign a negative vulnerability
modifier of 30% to preliminary vulnerability estimates (Table 14). As
for PDC vulnerabilities, reinforced concrete frame buildings that adhere
to locally recognised standards of earthquake design are not expected to
exhibit similar vulnerabilities to analogous buildings described in the
GEVES study (Spence et al., 2008) and so we similarly assign a negative
30% vulnerability modifier to previous estimates. Unfortunately, earth-
quake resistant RC frame buildings can be distinguished from non-
earthquake resistant RC frame buildings only with a comprehensive

Table 13
Defined wall and panel class vulnerabilities for failure under dynamic pressure loading from pyroclastic density currents, relevant to the surveyed building stock at Kanlaon and Fogo
volcanoes. Some (4% at Kanlaon and 27% at Fogo) of the buildings surveyed were plastered, which prohibited explicit identification of the construction type.

Wall/panel class Description Interdecile range (mean)

(% of surveyed) of failure pressure
(kPa)
Kanlaon Fogo

K_APDC (62) – Weak timber panels in a timber frame 1.0 to 2.5 (1.5)
K_BPDC (4) F_APDC (19) Unreinforced regular (cut stone/concrete) block masonry; 2.5 to 6.0 (4.0)
Reinforced concrete frame buildings without earthquake resistant design (non-aseismic)
– F_BPDC (28) Unreinforced rubble stone masonry with no frame 3.5 to 8.0 (5.5)
K_CPDC (30) F_CPDC (26) Reinforced concrete frame buildings with earthquake resistant design 5.0 to 10.5 (7.0)
118 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120

Table 14
Defined frame, wall and panel vulnerabilities for heavy or total damage under earthquake shaking at Kanlaon and Fogo volcanoes, relevant to the surveyed building stock. Some (4% at
Kanlaon and 27% at Fogo) of the buildings surveyed were plastered, which prohibited explicit identification of the construction type. Felt intensities follow the EMS-98 scale (Grünthal,
1998) and building failure is considered to occur with earthquake damage classes of D4 or greater, representing heavy or total structural failure. Vulnerability estimates are modified
from those created for the RADIUS (IDNDR, 1999) and GEVES (Spence et al., 2008) studies.

Frame, wall, panel class Description Probability (%) of heavy or total damage
(% of surveyed) at EMS-98 intensity:

Kanlaon Fogo V VI VII VIII

K_AEQ (4) F_AEQ (19) Unreinforced regular (cut stone/concrete) block masonry; 0.1 0.7 9 34
Reinforced concrete frame buildings without earthquake resistant design (non-aseismic)
– F_BEQ (28) Unreinforced rubble stone masonry 0.0 0.3 2 21
K_BEQ (30) F_CEQ (26) Reinforced concrete frame buildings with earthquake resistant design 0.0 0.0 1 3
K_CEQ (62) – Timber frame 0.0 0.2 0.7 3

structural investigation or with building records. Unreinforced regular
block masonry or RCF buildings without earthquake resistant design
(e.g. lower quality or uncompacted concrete mix, poorly placed or too
little steel reinforcements) are expected to show a lower vulnerability
to earthquakes than rubble stone masonry due to the reduced weight
and the regularity of block shapes. Failure probabilities determined spe-
cifically for building stock in developing countries during the RADIUS
study – Risk Assessment Tool for Diagnosis of Urban areas against Seis-
mic disasters (IDNDR, 1999) – offer a good indication of likely failure
probabilities for these separate building types (Table 14).
5.5. Vulnerability of agricultural support infrastructure
Assessment of the physical consequences of eruptive activity in de-
veloping regions, where Kanlaon, Fogo and many of the world's volca-
noes sit, are less well understood than in more developed regions. In
developing regions, livelihoods are often reliant upon agriculture,
which is supported by fertile soils in volcanic areas but also adversely af-
fected during eruption. In agricultural areas, key infrastructure that may
be affected by a future eruption includes that which supports on-farm
production (e.g. agricultural machinery, irrigation, power, transporta-
tion, storage), ensures efficient trading and exchange (e.g. telecommu-
nications, covered markets) and enables produce to move rapidly and
efficiently from farm to processing facilities and on to wholesalers
(e.g. roads, railways, transportation and bulk storage). Identification of
the major agricultural support infrastructure in exposed areas
surrounding Kanlaon and Fogo volcanoes was sourced predominantly
from fieldwork carried out as part of the MIA-VITA project. A key aspect
of the vulnerability of agricultural operations is their time-dependence
relative to harvesting. For example, fruits are most susceptible to dam-
age from ash falls just before and during harvest and their trade can
be indirectly affected by damage to the associated infrastructure. Even
relatively small thicknesses of ash (~ 1 mm: Table 7) can affect road
and rail transport and in turn the movement of crops to market, storage
or processing facilities. The timing and duration of an eruption within
the agricultural crop cycles will therefore be critical in forecasting erup-
tion consequences. Using data collected during our field studies, we
have been able to highlight key times throughout the year when an
eruption, even one depositing a few centimetres of ash, would have
the largest impact on the agricultural economy, and therefore liveli-
hoods, around Kanlaon and Fogo (Fig. 8). Combining such information
with hazard assessments that consider seasonal effects, e.g. changes in
wind conditions, increased rain, would provide a more accurate evalua-
tion of how the risk may vary across the year. The Kanlaon area will be
more vulnerable to eruptions over approximately seven months from
October to May as a result of the important sugarcane harvest, which
comprises 70% of cultivated land in the area, but also because this is
the dry season and remobilisation of ash may be a problem. The topo-
graphic high of Fogo volcano means that the most fertile part of the
Island lies to the northeast of the crater with coffee, wine and some
fruits being the major agricultural produce; however, prevailing wind
conditions are northeasterly so that higher ash hazard may be expected

Fig. 8. Key months for a) Kanlaon volcano, and b) Fogo volcanoes, where an eruption would have the greatest disruption for local livelihoods reliant upon agriculture. Two key crop types
for each volcano were chosen on the basis of their importance for the local economy; more detailed study for subsistence or small-scale farming would be valuable to extend this study.
Rice can be harvested year round and typically there are two to three harvests a year at Kanlaon; vulnerability will therefore be specific to each farm and perhaps each year.
 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120
in Chã and to the southwest of the volcano. The Fogo area, and in partic-
ular agricultural activities in Chã das Caldeiras, will be very vulnerable in
the month of April, which is also when rains are rare, but relatively re-
sistant from September to February.
Some discussion of the vulnerability of transport, water supply and
irrigation, power, telecommunications, processing facilities and market
and storage structures has been made in Jenkins and Spence (2010). Key
vulnerabilities include the road networks around both volcanoes, which
have several weaknesses that may cause problems during a volcanic
eruption. At Kanlaon, a national road crosses the southern flank of the
volcano exposed to ash fall, pyroclastic density currents, lahars and
lava flows. An eruption would threaten the major link between the
east and west of the island, potentially affecting agricultural trade and
the transport of sugarcane to mills and refineries. The condition of the
road is already poor in places and no alternative route exists. One of
the most vulnerable roads on Fogo crosses the east flank of the volcano,
connecting the larger communities of Mosteiros and Cove Figueira. This
road would be exposed to any flows exiting Chã das Caldeiras; however
due to the frequent brief-but-heavy rains between August and October,
bridges generally span deep, relatively narrow channels and are capable
of spanning large-volume fast-flowing rivers. Vulnerability to lahar is
therefore limited due to the large bridge clearance and relatively strong
construction and design. In both regions, it was clear that keeping infra-
structure functional is key to reducing impact on livelihoods and agri-
cultural activities. Emergency management planning and community-
based disaster risk reduction efforts should therefore consider the
vulnerability and interdependency of critical infrastructure (lifelines)
affected by volcanic hazards and outline potential risk reduction strate-
gies, e.g. mutual support agreements with infrastructure companies
outside the region of likely impact or methodologies and resources for
repeated clean-up.
6. Conclusions
Physical vulnerability assessment forms just one component of risk
assessment in a volcanic area. To carry out a volcanic risk assessment,
one would also need to consider the hazard (i.e. the intensity, spatial
distribution and frequency for each parameter in question), the number
and typology of elements that may be affected by the hazard (the expo-
sure) and the response of each element to the hazard intensity at that
site (the vulnerability). The ability to adequately quantify risk for an
area is thus dictated by the availability and quality of each component
of the assessment. Comprehensive volcanic hazard assessments should
therefore be accompanied by detailed field surveys that enable particu-
larly exposed and vulnerable areas to be highlighted, informing mitiga-
tion actions that could be undertaken prior to the event. For example,
given adequate time, risk can be reduced by limiting exposure through
land-use planning or decreasing the vulnerability of exposed elements
through structural retrofitting.
As part of the MIA-VITA project we have consolidated the experience
and expertise of previous volcanic vulnerability surveys and assess-
ments, predominantly in Europe, and used this to provide preliminary
vulnerability estimates that may act as a foundation for more refined
studies. Estimating physical vulnerability within a volcanic risk assess-
ment framework is a relatively recent study area compared to seismic
studies for example. As a result these initial estimates are subject to
varying degrees of uncertainty, which is a function of the number of
previous studies undertaken for each hazard and building type. For
example, volcanic earthquake vulnerability estimates borrow from the
extensive empirical, experimental and theoretical studies carried out
globally for tectonic earthquakes and as such are more robust than vul-
nerability estimates for pyroclastic density current impact. It is therefore
important that the uncertainties associated with each estimate are
propagated through any risk modelling or forecasting, ideally using
probabilistic techniques, to ensure that the full spectrum of possible
outcomes is considered. Given the complexity of volcanic phenomena
119
and difficulties in observing the dynamic parameters, any forecast of
likely building damage should also convey the uncertainties associated
with estimations of likely hazard intensity.
Field surveys at two case study volcanoes have shown that the volca-
nic hazard posing the greatest risk to the population depends to an im-
portant extent upon the construction methods and materials used. At
Kanlaon the buildings are most vulnerable to failure following PDC
and least vulnerable to failure from seismic shaking due to local building
types, in particular the predominance of timber frame buildings. Besides
volcanic hazards, areas around Kanlaon are vulnerable to building dam-
age from tectonic earthquakes and typhoons and construction has likely
considered these more frequent and well-known hazards above volca-
nic impacts. By contrast, at Fogo surveys found that buildings are most
vulnerable to failure following earthquakes and least vulnerable to fail-
ure from ash fall loading due to the high proportion of unreinforced ma-
sonry buildings with RC slab roofs. With this information, local
emergency management officials may identify more vulnerable build-
ings and building practices and, in concert with considering building
vulnerability to other hazards in the area such as typhoons or earth-
quakes, consider how best to mitigate the impact of a future event. For
example, a stronger roof structure well connected to walls in both direc-
tions can protect buildings against wind, PDC and ash loading but an in-
creased roof weight may mean the building is more vulnerable to
earthquake shaking.
Much less data are available regarding the vulnerability of infra-
structure to volcanic hazards. Up to this point, a greater priority has
been placed upon determining building vulnerability due to the poten-
tial impacts on occupants. A more qualitative discussion has therefore
been presented for infrastructure vulnerability compared to building
vulnerability. Potential infrastructure impacts from volcanic hazards
have been identified and categorised for key infrastructure sectors
where disruption may be expected to have the greatest consequences
for local livelihoods. Given knowledge of the specific materials, design
and configuration of each infrastructure type as well as their interde-
pendencies with other systems, bespoke relative or absolute vulnerabil-
ity estimates could be determined, but these were outside the scope of
the current project. Such estimates could not be easily transferred to im-
pact assessments in other areas.
To conclude, the preliminary vulnerability estimates presented here
represent starting points for local vulnerability assessments in volcanic
areas. They draw on available knowledge and expertise, which is not as
well developed or robust as for more frequent hazards such as earth-
quake or windstorm. Much of the recent advances in this area arise
from recent experimental and theoretical studies around Vesuvius in
Italy. It is therefore likely that future advances will continue to come,
to some extent, from experimental and theoretical studies; but it is
vital that empirical data collected following future damaging volcanic
events are collected and analysed wherever possible. Post-eruption
field missions should therefore always try to include quantitative stud-
ies of damage to the built environment, which as well as informing vul-
nerability assessments and associated mitigation strategies can be used
to constrain the calculation of hazard dynamics.
Acknowledgements
We would like to thank Angela Montes, Ben Tanatan and colleagues
at PHIVOLCS for field support at Kanlaon volcano. We would equally like
to thank Simon Day, Bruno Faria, Helder Ferreira, Antonio Gonçalves
and colleagues at the Laboratory for Civil Engineering of Cape Verde
for providing information and field support at Fogo volcano. The authors
are grateful for the hospitality and support of communities and local
authorities around both volcanoes, who generously donated their time
by providing information, data and offering their invaluable local expe-
rience and expertise. We also gratefully acknowledge the support of
Giulio Zuccaro and colleagues at the PLINIVS centre, University of
Naples, for contributing data and expertise from their studies of building
120 S.F. Jenkins et al. / Journal of Volcanology and Geothermal Research 276 (2014) 105–120

vulnerability around Vesuvius. This work was funded by the MIA- Marti, J., Spence, R., Calogero, E., Ordoñez, A., Felpeto, A., Baxter, P., 2008. Estimating build-
ing exposure and impact to volcanic hazards in Icod de los Vinos, Tenerife (Canary
VITA (EU FP7-ENV contract 211393) project, VOLDIES (ERC contract Islands). J. Volcanol. Geotherm. Res. 178 (3), 553–561.
228064) project and an AXA Research Fund fellowship. We thank Chiara Martinez-Villegas, M.L., Bornas, M.A.V., Abigania, M.I.T., Listanco, E.L., 2001. Geology and
Scaini, Barcelona Supercomputing Centre, and Christina Magill, Mac- geochemistry of Canlaon volcano, Negros Island, Philippines. J. Geol. Soc. Philipp. 56
(3 & 4), 163–183.
quarie University, for their insightful and helpful reviews that improved MIA-VITA project, 2009-2012. MItigate and Assess risk from Volcanic Impact on Terrain
the manuscript. and human Activities. http://miavita.brgm.fr.
Neri, M., Le Cozannet, G., Thierry, P., Bignami, C., Ruch, J., 2013. A method for multi-hazard
mapping in poorly known volcanic areas: an example from Kanlaon (Philippines).
References Nat. Hazards Earth Syst. Sci. 13 (8), 1929–1943.
Petrazzuoli, S.M., Zuccaro, G., 2004. Structural resistance of reinforced concrete buildings
Baxter, P.J., Boyle, R., Cole, P., Neri, A., Spence, R., Zuccaro, G., 2005. The impacts of pyro-
under pyroclastic flows: a study of the Vesuvian area. J. Volcanol. Geotherm. Res. 133
clastic surges on buildings at the eruption of the Soufrière Hills volcano, Montserrat.
(1–4), 353–367.
Bull. Volcanol. 67 (4), 292–313.
PHIVOLCS, 2003. Volcanic Hazards at Kanlaon Volcano. PHIVOLCS.
Biass, S., Frischknecht, C., Bonadonna, C., 2013. A fast GIS-based risk assessment for tephra
Pomonis, A., Spence, R., Baxter, P., 1999. Risk assessment of residential buildings for an
fallout: the example of Cotopaxi volcano, Ecuador. Part II: vulnerability and risk
eruption of Furnas Volcano, Sao Miguel, the Azores. J. Volcanol. Geotherm. Res. 92
assessment. Nat. Hazards 65 (1), 497–521.
(1–2), 107–131.
Blong, R.J., 1981. Some effects of tephra falls on buildings. In: Self, S., Sparks, R.S.J. (Eds.),
Rossetto, T., Iannou, I., Grant, D.N., 2013. Existing empirical vulnerability and fragility
Tephra Studies. Proceedings NATO Advanced Studies Institute, Laugarvatn and
functions: compendium and guide for selection. GEM Technical Report. The GEM
Reykjavik, June 18–29, 1980, pp. 405–420.
Foundation, Pavia.
Blong, R.J., 1984. Volcanic Hazards: A Sourcebook on the Effects of Eruptions. Academic
Schick, R., 1981. Source mechanism of volcanic earthquakes. Bull. Volcanol. 44 (3),
Press, Australia (424 pp.).
491–497.
Blong, R., 2000. Volcanic Hazards and Risk Management, Encyclopedia of Volcanoes.
Spence, R., Pomonis, A., Baxter, P., Coburn, A., White, M., Dayrit, M., Field Epidemiology
Academic Press 1215–1227.
Training Program Team, 1996. Building damage caused by the Mount Pinatubo erup-
Blong, R., 2003. Building damage in Rabaul, Papua New Guinea, 1994. Bull. Volcanol. 65
tion of June 15, 1991. In: Newhall, C.G., Punongbayan, R.S. (Eds.), Fire and Mud: Erup-
(1), 43–54.
tions and Lahars of Mount Pinatubo, Philippines. Philippine Institute of Volcanology
Casale, R., Fytikas, M., Sigvaldasson, G.E., 1998. The European laboratory volcanoes. Pro-
and Seismology. University of Washington Press, Seattle, pp. 1055–1061.
ceedings of the Second Workshop, Santorini, Greece, 2 to 4 May 1996. Directorate-
Spence, R., Brichieri-Colombi, N., Fawcett, W., Holdsworth, F., Zuccaro, G., Baxter, P., 2001.
General for Research and Innovation, European Commission, p. 650.
Vesuvius: volcanic hazards and structural mitigation strategies. Report for EXPLORIS.
Day, S.F., Faria, B.V.E., 2009. A geological hazard map of the island of Fogo showing broad
Spence, R., Baxter, P.J., Zuccaro, G., 2004. Building vulnerability and human casualty esti-
distribution of volcanic and other hazards. Draft Version.
mation for a pyroclastic flow: a model and its application to Vesuvius. J. Volcanol.
Day, S.J., Heleno da Silva, S.I.N., Fonseca, J.F.B.D., 1999. A past giant lateral collapse and
Geotherm. Res. 133 (1–4), 321–343.
present-day flank instability of Fogo, Cape Verde Islands. J. Volcanol. Geotherm. Res.
Spence, R., Kelman, I., Baxter, P., Zuccaro, G., Petrazzuoli, S., 2005. Residential building and
94 (1–4), 191–218.
occupant vulnerability to tephra fall. Nat. Hazards Earth Syst. Sci. 5, 477–494.
EXPLORIS project, 2004-2007. Explosive eruption risk and decision support for EU popu-
Spence, R., Kelman, I., Brown, A., Toyos, G., Purser, D., Baxter, P., 2007. Residential building
lations threatened by volcanoes. Funded by the European Union: Proposal no. EVR1-
and occupant vulnerability to pyroclastic density currents in explosive eruptions. Nat.
2001-00047, Contract no. EVR1-CT-2002-40026.
Hazards Earth Syst. Sci. 7, 219–230.
Faria, B., Fonseca, J.F.B.D., 2014. Investigating volcanic hazard in Cape Verde Islands
Spence, R., So, E., Jenny, S., Castella, H., Ewald, M., Booth, E., 2008. The Global Earthquake
through geophysical monitoring: network description and first results. Nat. Hazards
Vulnerability Estimation System (GEVES): an approach for earthquake risk assess-
Earth Syst. Sci. 14, 485–499.
ment for insurance applications. Bull. Earthq. Eng. 6 (3), 463–483.
Glasstone, S., Dolan, P.J., 1977. The Effects of Nuclear Weapons, United States Department
Valentine, G.A., 1998. Damage to structures by pyroclastic flows and surges, inferred from
of Defence and the United States Department of Energy. Third edition.
nuclear weapons effects. J. Volcanol. Geotherm. Res. 87 (1–4), 117–140.
European Macroseismic Scale 1998 (EMS-98). In: Grünthal, G. (Ed.), Centre Européen de
Wilson, T., Cole, J., Stewart, C., Cronin, S., Johnston, D., 2011. Ash storms: impacts of wind-
Géodynamique et de Séismologie, Luxembourg (99 pp.).
remobilised volcanic ash on rural communities and agriculture following the 1991
IDNDR, 1999. RADIUS: risk assessment tools for diagnosis of urban areas against seismic
Hudson eruption, southern Patagonia, Chile. Bull. Volcanol. 73 (3), 223–239.
disasters. Final Technical ReportInternational Decade for Natural Disaster Reduction
Wilson, T.M., Stewart, C., Sword-Daniels, V., Leonard, G.S., Johnston, D.M., Cole, J.W.,
(United Nations International Strategy for Disaster Reduction).
Wardman, J., Wilson, G., Barnard, S.T., 2012. Volcanic ash impacts on critical infra-
Jaiswal, K.S., Wald, D.J., D'Ayala, D., 2011. Developing empirical collapse vulnerability
structure. Phys. Chem. Earth A/B/C 45–46, 5–23.
functions for global building types. Earthquake Spectra 27 (3), 775–796.
Yamazaki, F., Muroa, O., 2000. Vulnerability functions for Japanese buildings based on
Jenkins, S., Spence, R., 2009a. Non-residential building types, vulnerabilities and mean
damage data from the 1995 Kobe earthquake. In: Elnashai.A. S, Antoniou, S. (Eds.),
damage ratios in the Vesuvius area. Technical Report D2 for Project SPeeD, Cambridge
Implications of Recent Earthquakes for Seismic Risk. Imperial College Press, London.
Architectural Research.
Zobin, V., 2001. Seismic hazard of volcanic activity. J. Volcanol. Geotherm. Res. 112, 1–14.
Jenkins, S., Spence, R., 2009b. Vulnerability curves for buildings and agriculture. Technical
Structural vulnerability to possible pyroclastic flows consequent to the eruption of the
Report D4.b for EU FP7-ENV Project MIA-VITA, Contract Number 211393.
volcano Vesuvius. In: Zuccaro, G. (Ed.), Final Report Vesuvius Project.
Jenkins, S., Spence, R., 2010. Physical vulnerabilities of buildings and agricultural support
Zuccaro, G., De Gregorio, D., 2013. Time and space dependency in impact damage evalu-
infrastructure: Kanlaon and Fogo volcanoes. Technical Report D4.d for EU FP7-ENV
ation of a sub-Plinian eruption at Mount Vesuvius. Nat. Hazards 68 (3), 1399–1423.
Project MIA-VITA, Contract Number 211393.
Zuccaro, G., Papa, F., 2002. Method of seismic vulnerability and exposure assessment at
Jenkins, S., Komorowski, J.C., Baxter, P., Spence, R., Picquout, A., Lavigne, F., Surono, 2013.
national scale — the Italian case. Proceedings XII European Conference on Earthquake
The Merapi 2010 eruption: an interdisciplinary impact assessment methodology for
Engineering.
studying pyroclastic density current dynamics. J. Volcanol. Geotherm. Res. 261,
Zuccaro, G., Cacace, F., Spence, R.J.S., Baxter, P.J., 2008. Impact of explosive eruption sce-
316–329.
narios at Vesuvius. J. Volcanol. Geotherm. Res. 178 (3), 416–453.
Macedonio, G., Costa, A., 2012. Rain effect on the load of tephra deposits. Nat. Hazards
Earth Syst. Sci. Brief. Commun. 12, 1229–1233.
Magill, C., Wilson, T., Okada, T., 2013. Observations of tephra fall impacts from the 2011
Shinmoedake eruption, Japan. Earth Planets Space 65 (6), 677–698.