longterm forecast of eruption style and size at campi flegrei caldera italy
TRANSCRIPT
-
8/12/2019 Longterm Forecast of Eruption Style and Size at Campi Flegrei Caldera Italy
1/12
Long-term forecast of eruption style and size at Campi Flegrei caldera (Italy)
Giovanni Orsi a,, Mauro Antonio Di Vito a, Jacopo Selva b, Warner Marzocchi c
a Istituto Nazionale di Geosica e Vulcanologia, Osservatorio Vesuviano, Via Diocleziano 328, 80124 Naples, Italyb Istituto Nazionale di Geosica e Vulcanologia, Sezione di Bologna, Via D. Creti 12, 40128 Bologna, Italyc Istituto Nazionale di Geosica e Vulcanologia, Sezione Roma 1, Via di Vigna Murata 605, 40123 Rome, Italy
a b s t r a c ta r t i c l e i n f o
Article history:
Received 12 May 2009
Received in revised form 20 July 2009Accepted 7 August 2009
Available online 12 September 2009
Editor: Y. Ricard
Keywords:
volcanic hazards assessment
eruption size
eruption style
Campi Flegrei caldera
The Campi Flegrei caldera is an active and restless volcano in the densely inhabited Neapolitan area of
southern Italy. Because of the very high value (lives, properties, infrastructures, etc.) exposed to potential
volcanic hazards, it is one of the areas at highest volcanic risk on Earth. In such a situation we have made an
attempt to contribute to assessment of its volcanic hazards by providing a quantitative probabilistic long-
term forecast of style and size of the next eruption.
We have evaluated the most relevant physical parameters of the 22 explosive eruptions of the Campi Flegrei
caldera over the past 5 ka. This time span has been taken as the reference period for volcanic hazards
assessment on the basis of the volcanic and deformation history of the caldera. The evaluated parameters
include dispersal, volume and density of the pyroclastic deposits, volume of erupted magma, total erupted
mass, and eruption magnitude. The obtained results permit a size classication of the explosive eruptions,
which are grouped into three sizes: small, medium, and large. On the basis of the reconstructed eruption
dynamics, we have considered a type event(s) representative of each size class and hypothesized the style of
the next event. An effusive eruption will likely generate a dome or very small lava ows, while an explosive
event of any size very probably will produce particles fallout and owage of pyroclastic density currents.
Using a Bayesian inference procedure, we have assigned a conditional probability of occurrence to each of
the eruption size classes. A small-size explosive eruption is the most likely event with a probability of about
60%; a large-size explosive eruption is the least likely event with a probability of about 4%; a medium-size
explosive eruption has a probability of occurrence of about 25%; an effusive eruption has about 11%probability of occurrence.
2009 Elsevier B.V. All rights reserved.
1. Introduction
Long-term volcanic hazards assessment and eruption forecasting,
especially for an active volcano within a densely inhabited area, are
among the most important tasks of modern volcanology (e.g.:Blong,
1984; Crandell et al., 1984; Tilling, 1989a,b, 2005; Sigurdsson et al.,
2000; Marti and Ernst, 2005). A comprehensive long-term hazard
estimation usually requires the denition of the portion of the entire
history of a given volcano to be taken as reference and the estimation
of vent location, type and size of the next eruption, and eruption
phenomena and their sequence during the event. In particular, the
probabilistic estimation of vent location, and type and size of the next
eruption is important as it permits a hazards zoning of the territory
and construction of volcanic hazards maps. Furthermore, for volcanic
elds and calderas, it takes into account the largest part of uncertainty
in volcanic hazards evaluation.
Long-term forecast of size and style of the next eruption has to be
based on detailed stratigraphical, sedimentological and volcanological
investigations of the eruptive products of a given volcano during the
portion of its history taken as reference. The results of these analyses
enable denition of eruption dynamics, particles transport and depo-
sition mechanisms, dispersal and volumeof the erupted products, and
nally volume, mass and discharge rate of the magma feeding each
past event. These parameters provide the basic for a quantitative
evaluation of the likely size classes of a future eruption, and of the
probabilities of occurrence of each of them.
The Campi Flegrei caldera (CFc) in the Neapolitan area is an ex-
cellent example of an active and restless volcano (Orsi et al., 1996).
Due to the very dense urbanization, it is one of the areas at highest
volcanic risk on Earth. The only volcanic hazard assessment based on
the entire set of available geological data for this structure was
recently attempted byOrsi et al. (2004)and later integrated byCosta
et al. (2009) and Selva et al. (submitted for publication-b).Orsi et al.
(2004)took the past 5 ka as the reference period for volcanic hazards
assessment based on the volcanic and deformation history of the
caldera. They subdivided the explosive eruptions that occurred in this
time span into three size classes, according to the area covered by the
Earth and Planetary Science Letters 287 (2009) 265276
Corresponding author. Tel.: +39 081 6108343; fax: +39 081 6108344.
E-mail address:[email protected](G. Orsi).
0012-821X/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsl.2009.08.013
Contents lists available at ScienceDirect
Earth and Planetary Science Letters
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e p s l
mailto:[email protected]://dx.doi.org/10.1016/j.epsl.2009.08.013http://www.sciencedirect.com/science/journal/0012821Xhttp://www.sciencedirect.com/science/journal/0012821Xhttp://dx.doi.org/10.1016/j.epsl.2009.08.013mailto:[email protected] -
8/12/2019 Longterm Forecast of Eruption Style and Size at Campi Flegrei Caldera Italy
2/12
pyroclastic deposits that were assumed to be representative of the
eruption magnitude.
The main goal of this paper is to improve the current volcanic
hazards assessment for CFc by providing a quantitative probabilistic
long-term forecast of style and size of the next eruption. To reach this
goal, we have improved the magnitude estimation of each of the
explosiveeruptions of the reference period through a detailed analysis
of their pyroclastic sequences, focusing on the evaluation of variable
physical parameters. The new magnitude estimations and their evolu-tion through time have been used as the primary data for a long-term
probabilistic forecastingof thesizeof thenext eruption at CFc, through
a methodology based on a Bayesian inference that accounts for both
prior models and past activity of the volcano during the reference
period (Marzocchi et al., 2004, 2008, submitted for publication).
2. The Campi Flegrei caldera
The CFc, together with SommaVesuvius strato-volcano and Ischia
volcano island, is one of the three active volcanoes of the Neapolitan
area (Orsi et al., 2003) (Fig. 1). The caldera system formed from two
major collapses related to the Campanian Ignimbrite (39 ka) and the
Neapolitan Yellow Tuff (NYT; 15 ka) eruptions (Orsi et al., 1992, 1996,
2004, and references therein).
In the past 15 ka, volcanism has been very intense within the NYT
caldera, which has been affected by an ongoing resurgence (Orsi et al.,
1996; Di Vito et al., 1999). The resurgence occurs through a simple-
shearing mechanism (Orsi et al., 1991) and has generated a maximum
upliftof about90 m of the LaStarzablock,composedof a marineterrace.
In recent decades, the NYT caldera has been affected by two major
bradyseismic events, in 196972 and 198284, that generated 170 and
180 cm of maximum ground deformation, respectively (Barberi et al.,
1984, 1989; Orsi et al., 1999a). These events were followed by an
irregular but net subsidence interrupted by several minor episodes
in 198889 (7 cm), 1994 (
-
8/12/2019 Longterm Forecast of Eruption Style and Size at Campi Flegrei Caldera Italy
3/12
by a deep reservoir with the top at about 8 km depth, and individual
eruptions were fed by small magma batches located at less than 8 km
depth. They have also shown that at present there is no evidence of
magma batches larger than 1 km3 at less than 4 km depth.
Location of the eruption vents is a good tracer of the structures
active through time. During the rst and second epochs, vents were
mostlylocated along thestructural boundary of theNYT caldera, while
during the third epoch the majority was in the northeastern sector of
the NYT caldera
oor, that is along normal faults cutting through thenortheastern portion of the resurgent block (Orsi et al., 1996, 1999a).
Duringthe secondperiod of quiescence (8.24.8 ka),priorto theonset
of the third epoch, the stress regime within the NYT caldera changed
(Orsi et al., 1996; Di Vito et al., 1999). This new regime still persists, as
shown by the dynamics of the recent unrest episodes, that have been
interpreted as short-term, transient events within the long-term
deformation related to resurgence (Orsi et al., 1999a,b).
Orsi et al. (2004)performed the rst volcanic hazards assessment
at the CFc that took into consideration all the geological, volcanolog-
ical, petrological and geophysical data available in literature. Their
work has been later integrated into more quantitative and probabi-
listic analyses of some aspects of volcanic hazards assessment such as
tephra fallout (Costa et al., 2009) andlocationof a futureeruptionvent
(Selva et al., submittedfor publication-b). Based upon thevolcanicand
deformation history of the past 15 ka, and the ongoing dynamics,
which persists since the last change in stress regime occurred prior to
onset of the third epoch (4.83.8 ka), Orsi et al. (2004) concluded that
the past 5 ka should be considered as the portion of the entire history
of the CFc system to be taken as reference for volcanic hazards
assessment and eruption size and style forecasting. They suggested
that time and location of the next eruption will be dictated by the
stress regime, although they could not rule out building up of over-
pressureunrelated to tectonic processesas a possible triggering factor.
According to the characteristics of the past volcanic events, the authors
concluded that a future explosive eruption will very likely include
magmatic and phreatomagmatic explosions with generation of sus-
tained columns and particles fallout, and pyroclastic density currents
(pdcs). Consideringthe shorttimeintervals at whicheruptionsfollowed
each other during the past eruptive epochs, Orsi et al. (2004) alsosuggested that a future eruption in the AgnanoSan Vito area, which is
within the area at highest probability of vent opening, could be therst
of a series of events closely spaced in time.
3. Physical parameters of the explosive eruptions of the past 5 ka
To date the only systematic attempt to evaluate the size of the past
explosive eruptions of the CFc was made by Orsi et al. (2004). The
authors, lackinga satisfactoryevaluationof thevolume of thepyroclastic
deposits emplaced over the past 15 ka, especially of those laid down by
pdcs, assumed the area covered by pyroclastic deposits as representa-
tive of the magnitude of the eruptions from which they were generated.
Based on this assumption and considering the area covered by pyro-clastic deposits thicker than 10 cm, they subdivided the explosive
events of the past 15 ka, into three classes: low-, medium-, and high-
magnitude eruptions. The deposits of the low-magnitude eruptions
covered areas
-
8/12/2019 Longterm Forecast of Eruption Style and Size at Campi Flegrei Caldera Italy
4/12
eld data andtheir linearinterpolation on semi-logarithmic plots, and
uses the ArcView GIS software for data analysis.
APD, the area covered by the entire pyroclastic sequence, was
calculated as the area covered by a sequence not less than 10 cm thick
on eld data, and then extrapolated to that covered by not less than
1 cm of deposit (Dell'Erba, 2003).VPDwas calculated as the volume of
the entire pyroclastic body, including both pyroclastic-fallout and pdc
deposits thicker than 1 cm (Pyle, 2000; Fierstein and Nathenson,
1992, 1993; Dell'Erba, 2003; Costa et al., 2009). Considering thatthe magmas feeding the explosive eruptions of the past 5 ka have
been dominantly trachytic in composition, the VDREwas evaluated by
assuming a magmatic density of 2500 kg/m3 (Romano et al., 2003).DPDwas evaluated on the basis of the density of each lithotype of the
pyroclastic sequence, resultingfrom eld and laboratory analyses, and
its percentage in volume of the entire deposit. Density values for
pyroclastic-fallout deposits with variable sedimentological character-
istics were already presented by Orsi et al. (2004). The eruption
magnitude, according toPyle (2000), was calculated by the formula
M= log10TEM; kg 7 1
The obtained results (Table 1) show a signicant variability for all
the calculated parameters. Excluding the AMS eruption, by far the
largest event of the past 5 ka, APD1cmvaries from 24 to 893 km2,VPDfrom 0.02 to 0.33 km3, VDREfrom 0.01 to 0.19 km
3, TEM from 0.4 to
4.7 kg1011, andMfrom 3.5 to 4.7.
The frequency distribution of physical parameters such asAPD, VPD,
VDRE, and TEM follows a power law, with the lower values for each of
them being the most frequent (Fig. 2). The distribution of theAPD1cmvalues, shows two breaks, which allow the entire set of data to be
subdividedin threeclasses. The small-area class includes 15 eruptions,
the medium-area class includes 6 eruptions, and the wide-area class
only includes one eruption. The same subdivision into three classes,
although not as obviously as for APD1 cm, can also be made using the
VPD, VDRE, and TEM frequency distributions. A gap in the magnitude
values around 4.3, which separatesevents of the small-area fromthose
of the medium-area class, can be taken as the boundary between the
low-magnitude and the medium-magnitude classes of eruptions. Alsoin this case, AMS is the only event in the high-magnitude class.
The variation through time of the measured physical parameters
(Fig. 3) sheds light on the past behavior of the volcanic system. Orsi
et al. (2004) presented and discussed the variation of theAPD10cm over
the past 15 ka. They found a striking similarity in the variation of this
parameter during the rst (159.5 ka) and third (4.83.8 ka) epoch of
activity, and grouped the explosive eruptionsin three size classes. Our
results conrm and extend those ofOrsi et al. (2004). During the third
epoch of activity, all the measured parameters roughly increased
untill the large AMS eruption and then decreased.
The Volcanic Explosivity Index (VEI), developed by Newhall and
Self (1982) by integrating a variety of parameters (Walker, 1973,
1980), is widely used to express the size of an eruption. The use of the
VEI index implies the assumption that magnitude, depending on thetotal erupted mass of magma, and intensity, related to the rate of
magma release and column height, are correlated. However this
correlation commonlyexists for largeeruptions, but not necessarily for
medium- and small-size events (Pyle, 2000). Large eruptions generate
high columns and produce large volumeof ejecta, while medium- and
small-size eruptions (especially phreatomagmatic events) can gener-
ate low columns and many pyroclastic currents. In the latter case, a
discrepancy arisesfrom VEIvalues calculated on thebasisof volumeof
ejecta (magnitude) or column height (intensity). For these reasons,
and because the volcanic eruptions of the past 5 ka have been domi-
nated by phreatomagmatic fragmentation processes, it does not seem
appropriate to apply the VEI index to categorize the past eruptions of
the CFc. Instead, magnitude is the most diagnostic parameter for such
a purpose. In fact, magnitude depends upon the volume of erupted
magma andpyroclastic deposits,i.e.,on theerupted mass, a parameter
even more useful to evaluate the impact of an eruption on the envi-
ronment. Considering the magnitude as well as the parameters upon
which it depends (Fig. 4) and their frequency distribution (Fig. 2), the
explosive eruptions of the past 5 ka can be subdivided according to a
threefold classication, which includes a small-, a large-, and a
medium-size class. Each class is characterized by a range of variability
of the measured parameters (Tables 1 and 2;Fig. 4). The large-size
class includes only theA
MS eruption; the medium-size class includes6 eruptions, namely Agnano 3, Paleo Astroni 2, Astroni 3, 4, 5, and 6;
and the remaining 15 eruptions fall within the small-size class.
According to thecharacteristics of theexplosive eruptionsof thepast
5 ka (de Vita et al., 1999; Di Vito et al., 1999, submitted for publication;
Isaia et al.,2004;Orsi et al.,2004; Costa et al.,2009; Tonarini et al., 2009),
we have singled out a representative type event(s) for each size class
(Table2). Syntheticdescriptionsof themaincharacteristicsof these type
eruptions are reported inAppendix A. The large-size class is character-
ized by AMS, the only high-magnitude eruption of the past 5 ka.
Eruptions of the medium-size class are represented by the Astroni 6
event. For the small-size class eruptions, Monte Nuovo and Averno 2
are suggested as two type events. Monte Nuovo, a tuff-cone forming
eruption, was dominated by phreatomagmatic explosions, while
Averno 2, a tuff ring-forming event, alternated magmatic phases, with
generation of sustained columns, to phreatomagmatic explosions.
4. Probability estimation of the size of the next eruption at
Campi Flegrei
The previouslydescribed data on the explosiveeruptions and those
available in literature on the effusive events, were used to estimate
the probability of each of the size-class eruptions at CFc for the next
eruption. For such a purpose, the eruptions have been grouped in 4
distinct size classes: size class 1 includes the effusive eruptions, while
size classes 2 through 4 include the dened small, medium and large
explosiveeruptions. By means of Bayesian inference, we have estimated
the conditional probability of an eruption of each of the 4 possible size
classes, given that an eruption occurs. Note that, following the consid-
erations reported above, the conditional probability of an eruption ofeach size class is assumed independent from its vent location.
In Bayesian inference, each probability is considered as a random
variable,characterized by its probability density function (e.g., Gelman
et al., 1995; Marzocchi et al., 2004, 2008, submitted for publication).
This process permits to take into account any kind of available infor-
mation (i.e., a priori and theoretical beliefs,models, expert opinions,as
wellas geological, historical and monitoring data). It alsoallows to deal
with aleatory and epistemic uncertainties in a structured, quantitative
and explicit fashion. Details on Bayesian inference and on the specic
choices adopted are reported inAppendix B. Here we only recall that
the prior/theoretical beliefs, models, expert opinions, etc., are sum-
marized in aprior distribution, whichis then combined withpast data.
The obtained combination is theposterior distribution and represents
the nal estimate of the procedure. The prior distribution is setthrough parameters such as theaverage probabilities to each class and
the equivalent number of data (see Eqs. (3) and (4) inAppendix B).
In our application to CFc, we have assigned an average probability
of 0.05 to the effusive eruptions size class (1) and of 0.95 to the
explosive eruption size classes (2 through 4). This choice is quite
arbitrary, butit roughly reectsthe effusiveversus explosiveeruptions
ratio over the past 15 ka (Di Vito et al., 1999; Orsi et al., 2004). Within
the explosive eruption size classes (2 through 4), we have set prior
probabilitiesaccording to a powerlaw estimatedby considering all the
historically known volcanic eruptions (Simkin and Siebert, 1994), in
analogy with Marzocchi et al. (2004) for Mt. Vesuvius. Therefore,
these choices lead to best guesses of theprior distributionof 0.05, 0.79,
0.13, and 0.03, respectively. The equivalent number of data has been
set to 1, i.e., the variance (epistemic uncertainty) is the maximum
268 G. Orsi et al. / Earth and Planetary Science Letters 287 (2009) 265276
-
8/12/2019 Longterm Forecast of Eruption Style and Size at Campi Flegrei Caldera Italy
5/12
Fig. 2.Frequency distribution of parameters calculated for the 22 explosive eruptions of Campi Flegrei caldera over the past 5 ka. (Data of Table 1).
269G. Orsi et al. / Earth and Planetary Science Letters 287 (2009) 265276
-
8/12/2019 Longterm Forecast of Eruption Style and Size at Campi Flegrei Caldera Italy
6/12
value allowed (seeMarzocchi et al., 2004, submitted for publication).
As regards past data, outof the25 eruptionsthatoccurredoverthe past
5 ka, 3 are effusive, while 15 are small-size, 7 medium-size, and 1
large-size explosive events.
The results (posterior probability distribution) obtained for the
four size classes are reported inTable 3. The probability distribution
for each size class is described through parameters such as best guess
probabilities (average) and condence intervals (10th, 50th and 90thpercentiles).
It is noteworthy that such probability distributions, despite being
an important result by themselves, are a basic component for both
achieving a comprehensive probabilistic volcanic hazards assessment
(Marzocchi et al., submitted for publication; Selva et al., submitted for
publication-a), and establishing quantitative decision-making proce-
dures (Marzocchi and Woo, 2007, 2009; Woo, 2008).
5. Discussion and nal remarks
Oneof thebasicassumptionsof this work is that the characteristics
of all the eruptions of the CFc over the past 5 ka and the physical
parameters measured for all the 22 explosive events are indicative of
the past behavior of the system and form a solid scientic basis for a
long-term forecast of style and size of a future eruption. We have
estimated the most relevant physical parameters of the explosive
eruptions, such as area covered by pyroclastic deposits thicker than 1
and 10 cm, volume and density of the pyroclastic deposits, volume of
erupted magma, total erupted mass, and eruption magnitude. The
obtained results have enabled us to perform a size classication of the
past eruptions and a probabilistic forecast of style and size of the next
event. According to size, the 22 explosive eruptions of the past 5 kacan be grouped in a threefold classication, which includes a small-, a
medium-, and a large-size class (Fig. 4). On the basis of the charac-
teristics of these eruptions, we have singled out a type event(s) for
each size class. AMS is the only large-size eruption and therefore the
type event for this class,whilewe have identied Astroni 6 as the type
event for the medium-size class. For the small-size class, we have
identied the Monte Nuovo and Averno 2 eruptions as type events,
because, although of similar size they have had different dynamics
and impact on the environment.
This understanding of the dynamics of the past eruptions permits
some general hypotheses on the style of the next event. An effusive
eruption will be very likely dominated by extrusion of viscous lavas,
which will generate a dome or very small lava ows. An explosive
eruption of any size very probably will be characterized by alternation
Fig. 3.Variation through time of some parameters calculated for the 22 explosive eruptions of Campi Flegrei caldera over the past 5 ka. Inset gures expand the time scale to better
show details.
270 G. Orsi et al. / Earth and Planetary Science Letters 287 (2009) 265276
-
8/12/2019 Longterm Forecast of Eruption Style and Size at Campi Flegrei Caldera Italy
7/12
of magmatic and phreatomagmatic phases, which will produce
particles fallout and owage of pdcs. At present it is impossible to
hypothesize a time sequence of the various eruption dynamics and
related phenomena during the course of a future event. Finally, a
sequence of eruptions following each other in the same vent area at
short (years or tens of years) time intervals cannot be ruled out.
A probability of future occurrence has been assigned to each of the
eruption size classes by means of a Bayesian inference procedure. Toeach class has been attributed an a priori distribution based on a
power law distribution typical for the global catalogue. This prior
distribution has been then updated through the likelihood containing
all the information of the eruptions that occurred over the past 5 ka.
The obtained results (Fig. 5) show that the most likely event to occur
is an explosive eruption of the small-size class with a conditional
probability of about 60%, the least likely event to occur is one of the
large-size class with a conditional probability of about 4%, while an
explosive eruption of the medium-size class has a conditional
probability of occurrence of about 25%. The conditional probabilityof occurrence of an effusive eruption is about 11%.
Fig. 4.Area1cmversus selected parameters calculated for the 22 explosive eruptions of Campi Flegrei caldera over the past 5 ka.
Table 2
Physical parameters variation and type eruption(s) for each of the dened size classes
for explosive events.
Size A1cm(km2)
VTephra(km3)
VDRE(km3)
TEM
(kg1011)
Magnitude Type
esruption(s)
Large >500 >0.40 > 0.3 > 5 >5 Agnano
Monte Spina
Medium 5001000 0.150.40 0.10.3 25 4.35 Astroni 6
Small 0500 00.15 00.1 02
-
8/12/2019 Longterm Forecast of Eruption Style and Size at Campi Flegrei Caldera Italy
8/12
The results presented in this paper increase the current knowledge
on the past behavior of the CFc system and supply both the scientic
community and the civil authorities with an important element in their
effortsto assess volcanic hazardsand to develop volcanic risk mitigation
measures. Furthermore, the low probability value of the large-size class
seems to suggest that risk mitigation measures based on the medium-
size class event could be the most reasonable. On the other hand, the
probabilities of all size classes should concur to dene a comprehensive
evaluation of volcanic hazards and risk (Marzocchi et al., submitted
for publication; Selva et al., submitted for publication-a).
Acknowledgements
The research has been carried out in the framework of the INGV-
DPC joint projects on the Italian active volcanoes. F. Dell'Erba is
thanked for the kind help in eld work and data processing. Two
anonymous reviewers are thanked for improving a rst version of the
manuscript.
Appendix A
A.1. AgnanoMonte Spina eruption
The AgnanoMonte Spina is thehighest-magnitudeeruptionof the
CFc over the past 5 ka (de Vita et al., 1999 and references therein)(Fig. 6). The measured physical parameters of this eruption are
reported inTables 1 and 2.de Vita et al. (1999)presented a detailed
reconstruction of the entire erupted sequence, dated the eruption
at 4,10050 years bp, and well constrained shape and size of the
volcano-tectonic collapse that accompanied the event. The erupted
sequence includes beds generated by contrasting fragmentation and
transportation dynamics. Magmatic Plinian/sub-Plinian fallout depos-
its commonly alternate with base-surge beds of phreatomagmatic
origin. Moreover, during some eruption phases, the contrasting erup-
tive dynamics were almost contemporaneous (Dellino et al., 2001,
2004). Variation of lithological features and occurrence of a well-
dened erosional unconformity allowed the authors to subdivide
the whole sequence into six Members named A through F, each
representing the deposit of one eruption phase. Members were fur-
ther subdivided into sub-Members, based upon sedimentological
characteristics.
The eruption began with phreatomagmatic explosions that pro-
duced a highly expanded ash cloud and deposition of an ash bed,
followed by a magmatic explosion that generated a low (about 5 km)
eruption column from which was deposited the upper fallout bed of
sub-Member A1. Collapse of this column and phreatomagmatic explo-
sions generated the sequence of ash-to-lapilli size surge- and fallout-
beds of sub-Member A2.
Formationof a pulsating eruption columnmarked the beginningof
the second phase of the eruption. This column reached a maximum
height of about 23 km, was dispersed towards the east, and deposited
the fallout layer B1 up to a distance of 45 km. Its destabilization
generated the basal pyroclastic surges of sub-Member B2, while its
Fig. 5.Relative proportions of conditional probability for all the events likely to occur
(best guess values) at Campi Flegrei caldera.
Fig. 6. Areal distribution of AgnanoMonte Spina deposits; the dashed line delimits the area of the pyroclastic density current dispersion, the solid lines encompass the 10 cm
isopachs of the main fallout deposits, the asterisk indicates the vent area. Modied afterde Vita et al. (1999) and Costa et al. (2009) .
272 G. Orsi et al. / Earth and Planetary Science Letters 287 (2009) 265276
-
8/12/2019 Longterm Forecast of Eruption Style and Size at Campi Flegrei Caldera Italy
9/12
subsequent waning produced owage of high particle-concentration
pdcs and a lag breccia within the Agnano area. At this stage, a network
of fractures, foreshadowing a volcano-tectonic collapse, likely formed
and became the site of scattered vents. Opening of fractures allowed
ground water to access different parts of the magma reservoir.
Therefore, the ensuing explosions were triggered by magma/water
interaction processes of variable efciency and generated the pdc
deposits of the upper part of sub-Member B2.
The second phase was followed by a pause in the eruption duringwhich heavy rains caused diffuse erosion along steep slopes and then
thene ash suspended in the atmosphere fell on the ground to form
Member C, a bed with abundant mud lumps.
An increase in fracturing of the roof rocks of the magmatic feeding
reservoir, probably caused lowering of the lithostatic pressure with
consequent magma volatile exsolution and migration towards the
surface. Interaction of the rising magma with the overheated geother-
mal system caused its ashing and resumption of the eruption, with
phreatomagmatic explosions and deposition of the pyroclastic-surge
beds at the base of Member D (sub-Member D0). The subsequent
explosions were magmatic and generated a Plinian column that reached
a maximum height of 27 km, was dispersed towards the northeast,
and deposited a fallout sequence (sub-Member D1) up to a distance of
45 km. The main episode of volcano-tectonic collapse took place during
this phase. Related to this episode, alternation of phreatomagmatic ex-
plosions and/oreruption column collapse generatedpdcs that deposited
the sequence of pyroclastic deposits characterized by different facies,
according to distance from thevent and to substrate morphology of sub-
Member D2.
The latest stages of volcano-tectonic collapse were accompanied
by vent migration towards the northwestern part of the collapsed
area. Variation of the structural setting of the Agnano area allowed a
large amount of ground water and geothermal uids to access the
reservoir. Phreatomagmatic explosions produced an expanded cloud,
which deposited the sequence of base surge beds of sub-Member E1,
and were followed by a short-lived magmatic eruption column, which
generated sub-Member E2, and by phreatomagmatic explosions that
generated the sequence of pdc and minor fallout deposits of sub-
Member E3. The nal phase of the eruption was characterized by
phreatomagmatic explosions producing ash deposits (Member F).
Accumulation ofne particles suspended in the atmosphere, favored
also by rainfall, marked the end of the eruption.
A.2. Astroni volcano
The Astroni volcano (Tonarini et al., 2009, and references therein)formed between 4.1 and 3.8 ka, in the northwestern portionof the area
affected by the volcano-tectonic collapse related to the AMS eruption
(de Vita et al., 1999)(Fig. 7). It is the best example of a long-lasting
activity at CFc, as it was constructed through at least 7 eruptions of
variable magnitude, following each other at short time intervals within
the same vent area. The deposits of the 7 eruptions, named Unit 1
through 7 (Fig. 7), are separated by erosional unconformities or thin
paleosols containing evidence of Eneolithic settlements (Marzocchella,
1998), suggestingtimebreaks between eruptions. Theeruptiveunitsare
composed ofne- to coarse-ash pdc deposits and subordinate fallout
layers. These eruptions, whose vents, although all located within the
present crater, migrated through time, were characterized by phreato-
magmatic with minor magmatic explosions, only two events ended
with low-energy explosions and lava extrusion. Unit 6, whose vent was
located in the southeastern sector of the present crater, comprisesne-
to coarse-ash beds with intercalated coarse pumice lenses and beds. A
coarse pumice Plinian layer forms the base, while more pumice fallout
beds occur in theupper portion of the sequence. The Plinianlayer is the
only such layer of the entire Astronisequence, and wasdeposited by an
eastward-dispersed eruption column, which reached a maximum
height of about 20 km. The measured physical parameters of this
eruption are reported inTables 1 and 2.
On the basis of stratigraphical and sedimentological characteristics
of the entire Astroni sequence,Isaia et al. (2004)have reconstructed
the history of the volcano, which is here summarized. The rst erup-
tion was dominated by phreatomagmatic with subordinate magmatic
explosions and was followed by a pause. Volcanism resumed with the
Fig. 7.Areal distribution of the deposits of the 7 eruptions of Astroni volcano. The solid line 6 encompasses the entire sequence of the Astroni Unit 6. The dashed line 6 is the 10 cm
isopach of the coarse fallout deposit of Unit 6. Modied afterIsaia et al. (2004) and Costa et al. (2009) .
273G. Orsi et al. / Earth and Planetary Science Letters 287 (2009) 265276
-
8/12/2019 Longterm Forecast of Eruption Style and Size at Campi Flegrei Caldera Italy
10/12
second and lowest energy eruption of the volcano, which was mostly
phreatomagmatic and generated contemporaneous particles fallout
and pdcs owage. A subsequent short pause was interrupted by the
third and largest eruption of the volcano, which was characterized
by alternation of phreatomagmatic and magmatic explosions, with
dominant phreatomagmatic events only during its nal stages.
Another quiescence preceded the fourth eruption, which was very
similar in dynamics to the previous one, although more energetic. A
newpause in the activity took place until volcanism resumed with thefth eruption, which was explosive and ended with lava extrusion.
Another pause was interrupted by the eruption that produced
Unit 6. This eruption began with magmatic explosions generating a
sustained column, the only Plinian column in the entire life of the
volcano, and continued with prevailing phreatomagmatic and sub-
ordinate magmatic explosions. In the nal phase of the eruption,
magmatic explosions prevailed and formed low columns that
deposited fallout beds. This sixth eruption likely was followed by a
short pause, interrupted by the last eruption of the volcano, which
began with an explosivephaseand ended with low-energy explosions
and lava effusions within the crater area.
A.3. Averno 2 eruption
The Averno 2 eruption occurred 3,700 50 years bp (Alessio et al.,
1971) and was characterized by magmatic and phreatomagmatic
explosions that produced a sequence of pyroclastic-fall and pdc
deposits. The whole sequence has been subdivided into three members
named A through C, from base upsection(Costaet al.,2009; DiVitoet al.,
submitted for publication) (Fig. 8). Isopachs and isopleths maps of
fallout deposits, as well as dispersal of ballistic fragments and areal
distributionof facies of pyroclasticdensity current deposits, suggest that
the vent migrated from southwest to northeast, likely along a NESW
trending fault system. The measured physical parameters of this
eruption are reported inTables 1 and 2.
The eruption began with prevailing magmatic and subordinate
phreatomagmatic explosions, which generated the sequence of
Member A, distributed over an area of about 37 km2. The magmatic
explosions produced at least six eruption columns, which reached amaximum height of 10 km and deposited fallout beds, dispersed
towards the southwest. The phreatomagmatic explosions produced
pdcs that deposited thin ash layers around the vent area.
The second phase of the eruption was characterized by phreato-
magmatic with subordinate magmatic explosions. The latter mostly
occurred at the beginning and at the end of the eruption phase. The
activity of this phase generated pdcs with subordinate low eruption
columns, which deposited Member B, distributed preferentially north
of the volcano over an area of about 34 km2. This member includes a
sequence of cohesive, massive to cross-laminated, generally ne-ash
surge beds, containing abundant accretionary lapilli and minor lapilli-
sized pumice fragments. Fallout deposits are concentrated in the
lower and upper parts of the sequence.
The third and last phase of the Averno 2 eruption also generated
explosions driven mainlyby variably efcient water-magmainteraction
and subordinately by magmatic fragmentation. The former mechanismproduced pdcs, while the latter generated low eruption columns. The
deposits of this phase formed Member C, distributed over an area of
about 12 km2, preferentially north of the vent area. The pdcs deposited
a sequence of cross-laminated to plane-parallel ash-surge beds and
bedsets containing lapilli-sized pumice fragments. This sequence in-
cludes a few thin fallout layers deposited by the low eruption columns.
The surge beds are sometimes overlain by massive ne-ash, accretion-
ary lapilli-rich fallout beds, and underlain by lenses of traction carpet or
by erosional unconformities. On the northernank of thevolcano, large
ballistic blocks and bombs form impact sags on surge deposits.
A.4. Monte Nuovo eruption
The Monte Nuovo eruption, the most recent event of the CFc, has
been reconstructed through both geological, volcanological and petro-
logical investigations, and analyses of historical documents (Di Vito
et al., manuscriptin preparation)(Fig.9). Theeruption,which lastedone
week and was fed by three vents, was characterized by three phases
separated by pauses in the activity. The main vent (MV) was located in
the present crater of the Monte Nuovo tuff cone, whereas two minor
vents were along the southern (SV) and northeastern (NEV) slopes
of the cone. The entire sequence of deposits has been subdivided in 5
members named A through E, from base upsection. The measured
physical parameters of the eruption are reported inTables 1 and 2.
The eruption began on September 29, 1538, at 7 p.m., and its rst
phase, which was the main phase of the entire event, lasted two days,
until the night of September 30. This phase generated almost contin-
uous phreatomagmatic with subordinate magmatic explosions, pro-
ducing pdcs and minor short-lived, low eruption columns, whichdeposited members A and B. Member A, composed of a sequence of
plane-parallel to undulatedne- to coarse-ash beds forms the largest
part of the Monte Nuovo tuff cone, and was erupted in about 12 hours
through the MV. Phreatomagmatic explosions at the SV produced
mainly pyroclastic density currents that deposited Member B. This
member, composed of wavy ne-ash deposits containing coarse
pumice and lithic fragments, is distributed only in the southern sector
of Monte Nuovo. Strombolian explosions at the SV and NEV deposited
Fig. 8. Areal distribution of the deposits of Averno 2 eruption. Curve a =cumulative
dispersal of the fallout deposits of Member A; curve b =cumulative pyroclastic density
current deposits of Member B; curvec=cumulative pyroclastic density current depos-
its of Member C. Modied afterCosta et al. (2009).
Fig. 9.Areal distribution of the deposits of Monte Nuovo eruption. Curve a =cumula-
tive dispersal of fallout deposits; curve b =cumulative pyroclastic density current
deposits. Modied afterOrsi et al. (2004).
274 G. Orsi et al. / Earth and Planetary Science Letters 287 (2009) 265276
-
8/12/2019 Longterm Forecast of Eruption Style and Size at Campi Flegrei Caldera Italy
11/12
the sequence of coarse scoria fallout deposits dispersed over narrow
areas around the eruption vents, which form Member C. This activity
marked the end of the rst phase and was followed by a pause that
lasted two days.
The eruption resumed on October 3 at 4 p.m. and lasted until the
next night. This second phase of the eruption was characterized by a
discontinuous sequence of low-energy phreatomagmatic and mag-
matic explosions at the MV, which deposited the sequence of Member
D. The phreatomagmatic explosions produced laminated to massiveash surge deposits, while the magmatic events produced coarse
pumice and scoria fallout deposits.
Followinga briefpause, thethird phaseof theeruption beganat 4 p.
m., October 6, and lasted few hours. This last phase was characterised
by low-energy magmatic explosions from the MV. This activity was
likelycharacterized by theexplosionof a small dome grownduring the
preceding pause, followed by minor, very low energy events at the
crater of the MV, which produced Member E. This member includes
fallout layers composed of dense to low-vesiculated, angular dark
clasts. During this phase, 24 people died while climbing the slopes of
the newly formed cone at the end of the second pause in the activity.
Appendix B
Bayesian inference is the process oftting a probability model to a
set of data andsummarizingthe results by a probability distribution of
the parameters of the model. To assess the conditional probability iof an eruption of size class i, given the set of data y and that an
eruption occurs, a probabilistic model to link the variables must be
used, i.e., the Bayes' Theorem:
i= ijy= ipriory ji=y 2
where the index i is relative to the size class of interest; [i|y] is the
posterior distribution, given the set of data y; [i]prior is the prior
distributioncontainingthe current knowledgebasedonly on theoretical
models, beliefs and/or expert opinion;[y|i] is the sampling distribution
(theso-called likelihood function), that is the probability distribution of
observing the datay given a specic probability of occurrence in the i-thevent (i); [y] is a normalizing factor accounting for the total probability
of observing datay.
The choice of the functional form of both prior distribution and
likelihood function is the core of the Bayesian inference, and requires
some physical and statistical assumptions. In the application pre-
sented here, we have modeled the prior distribution [i]prior with a
Dirichlet distribution with 4 possible outcomes, i.e., Di4(.) (see
Marzocchi et al., 2004, 2008, submitted for publication).This choice
simply assumes that size classes represent a set of mutually exclusive
and exhaustive events. The likelihood function [y|i] is modeled by a
multinomial distribution (seeMarzocchi et al., 2004, 2008, submitted
for publication), assuming that any datum in the y data set is inde-
pendent fromthe others. Withthese choices,the posteriordistribution
is still a Dirichlet distribution because the Dirichlet is the conjugatedistribution in the multinomial model (Gelman et al., 1995). This
choice is subjective and other distributions can be used, such as the
Gaussian distribution for the logistic transformation of the probability.
In practice, the differences associated with the use of reasonable
distributions are not typically signicant; for this reason, the Dirichlet
(or the Beta) distribution is the most used in practical applications
(see, e.g.,Gelman et al., 1995).
The posterior Dirichlet distribution can be expressed as
i= Di41;2;3;4 3
where
k =
k
+ 3+ yk k= 1;
; 4 4
In Eq. (4) the unknowns iand , andyiconcern the prior model
and the past data, respectively. The central value i represents the
best guess (aleatory uncertainty) of the prior distribution, the equiva-
lent number of data , being a measure of the dispersion around i,
represents our condence on the best guess (epistemic uncertainty).yi is the number of past eruptions in size class i. A more detailed
discussion about this formulation can be found in Marzocchi et al.
(2008, 2009)and in the references therein.
References
Alessio, M., Bella, F., Improta, S., Belluomini, G., Cortesi, C., Turi, B., 1971. University ofRome carbon-14 dates IX. Radiocarbon 13, 395411.
Arienzo, I., Moretti, R., Civetta, L., Orsi, G., Papale, P., submitted for publication. Thefeeding system of the Agnano-Monte Spina eruption (Campi Flegrei, Italy):dragging the past into present activity and future scenarios. Chem. Geol.
Barberi, F., Corrado, G., Innocenti, F., Luongo, G., 1984. Phlegraeanelds 19821984:brief chronicle of a volcano emergency in a densely populated area. Bull. Volcanol.47, 175185.
Barberi, F., Carapezza, M., Innocenti, F., Luongo, G., Santacroce, R., 1989. The problem ofvolcanic unrest: the Phlegraean Fields case history. Atti Conv. Lincei 80, 387405.
Blong, R.J., 1984. Volcanic hazards: a sourcebook of the effects of eruptions. AcademicPress, Sydney.
Costa, A., Dell'Erba, F., Di Vito, M., Isaia, R., Macedonio, G., Orsi, G., Pfeiffer, T., 2009.Tephra fallout hazard assessmentat the CampiFlegrei caldera (Italy). Bull.Volcanol.71, 259273.
Crandell, D.R., Booth, B., Kusumadinata, D., Shimozuru, D., Walker, G.P.L., Westerdamp,
D., 1984. Source Book for Volcanic-Hazards Zonation. UNESCO, Paris.D'Antonio, M.,Civetta,L., Orsi, G.,Pappalardo,L., Piochi,M., Carandente, A.,de Vita, S.,DiVito, M.A., Isaia, R., Southon, J., 1999. The present state of the magmatic system ofthe Campi Flegrei caldera based on the reconstruction of its behaviour in the past12 ka. J. Volcanol. Geotherm. Res. 91, 247268.
D'Antonio, M., Tonarini, S., Arienzo, I., Civetta, L., Di Renzo, V., 2007. Components andprocesses in the magma genesis of the Phlegrean Volcanic District (Southern Italy).In: Beccaluva, L., Bianchini, G., Wilson, M. (Eds.), Cenozoic volcanism in theMediterranean area: GSA, Sp. Pap. 418, Boulder, pp. 203220.
D'Oriano, C., Poggianti, E., Bertagnini, A., Cioni, R., Landi, P., Polacci, M., Rosi, M., 2005.Changes in eruptive style during the A.D. 1538 Monte Nuovo eruption (PhlegreanFields, Italy): the role of syn-eruptive crystallization. Bull. Volcanol. 67, 601621.
Dell'Erba, F., 2003. Denizione di parametric sici di alcune eruzioni esplosive dellacaldera dei Campi Flegrei negli ultimi 15 ka: implicazioni per la valutazione dellapericolosit vulcanica. PhD Thesis, University of Bari, Italy.
Dellino, P., Isaia, R., La Volpe, L., Orsi, G., 2001. Statistical analysis of textural data fromcomplex pyroclastic sequences: implication for fragmentation processes of theAgnanoMonte Spina Tephra (4.1 ka), Phlegraean Fields southern Italy. Bull.Volcanol. 63, 443461.
Dellino, P.,Isaia, R.,La Volpe,L., Orsi, G.,2004. Interaction between particlestransportedby fallout and surge in the deposits of the Agnano-Monte Spina eruption (CampiFlegrei, Southern Italy). J. Volcanol. Geotherm. Res. 133, 193210.
de Vita, S., Orsi, G., Civetta, L., Carandente, A., D'Antonio, M., Di Cesare, T., Di Vito, M.,Fisher, R.V., Isaia, R., Marotta, E., Ort, M., Pappalardo, L., Southon, J., 1999. TheAgnanoMonte Spina eruption in the densely populated, restless Campi Flegreicaldera (Italy). J. Volcanol. Geotherm. Res. 91, 269301.
Di Vito, M.A., Isaia, R., Orsi, G., Southon, J., de Vita, S., D'antonio, M., Pappalardo, L.,Piochi, M., 1999. Volcanism and deformation in the past 12 ka at the Campi Flegreicaldera (Italy). J. Volcanol. Geotherm. Res. 91, 221246.
Di Vito, M.A., Arienzo, I., Braia, G., Civetta, L., D Antonio, M., Di Renzo, V., Orsi, G.,submitted for publication. The Averno 2 ssure eruption: a recent small-sizeexplosive event at the Campi Flegrei caldera. Bull. Volcanol.
Fierstein, J., Nathenson, M., 1992. Another look at the calculation of fallout tephravolumes. Bull. Volcanol. 54, 156167.
Fierstein, J., Nathenson, M., 1993. Reply to comment by WI Rose. Bull. Volcanol. 55,156378.
Gelman, A., Carlin, J.B., Stern, H.S., Rubin, D.B., 1995. Bayesian data analysis. CRC, BocaRaton, Florida.
Isaia, R., D'Antonio, M., Dell'Erba, F., Di Vito, M., Orsi, G., 2004. The Astroni volcano: theonly example of close eruptions within the same vent area in the recent history ofthe Campi Flegrei caldera (Italy). J. Volcanol. Geotherm. Res. 133, 171192.
Jurado-Chichay, Z., Walker, G.P.L., 2001. The intensity and magnitude of the Mangaonesubgroup plinian eruptions from Okataina Volcanic Centre, New Zealand. J. Volcanol.Geotherm. Res. 111, 219237.
Legros, F., 2000. Minimum volume of tephra fallout deposit estimated from a singleisopach. J. Volcanol. Geotherm. Res. 96, 2532.
Mangiacapra, A., Moretti, R., Rutherford, M., Civetta, L., Orsi, G., Papale, P., 2008. Thedeep magmatic systemof theCampiFlegrei caldera (Italy). J. Geophys. Res. Lett. 35,L21304.
Marti, J., Ernst, G.G.J., 2005. Volcanoes and the environment. Cambridge UniversityPress, Cambridge.
Marzocchella, A.,1998. Tutela Archelogicae Preistorianella Pianura Campana. In:Guzzo,P.G., Peroni, R. (Eds.), Archeologia e Vulcanologia in Campania. Arte Tipograca,Naples, pp. 97133.
Marzocchi, W., Sandri, L., Gasparini, P., Newhall, C., Boschi, E., 2004. Quantifyingprobabilities of volcanic events: the example of volcanic hazard at Mt. Vesuvius.
J. Geophys. Res. 109, B11201.
275G. Orsi et al. / Earth and Planetary Science Letters 287 (2009) 265276
-
8/12/2019 Longterm Forecast of Eruption Style and Size at Campi Flegrei Caldera Italy
12/12
Marzocchi, W., Sandri, L., Selva,J., 2008. BET_EF: a probabilistic tool for long- and short-term eruption forecasting. Bull. Volcanol. 70, 623632.
Marzocchi, W., Sandri, L., Selva J., submitted forpublication. BET VH:a probabilistic toolfor long-time volcanic hazard assessment. Bull. Volcanol.
Marzocchi, W., Woo, G., 2007. Probabilistic eruption forecasting and the call for anevacuation. Geophys. Res. Lett. 34, L22310.
Marzocchi, W., Woo, G., 2009. Principles of volcanic risk metrics: theory and the casestudy of Mt. Vesuvius and Campi Flegrei (Italy). J. Geophys. Res. 114, B03213.
Newhall, C.G., Self, S., 1982. The Volcanic Explosivity Index (VEI): an estimate ofexplosive magnitude for historical volcanism. J. Geophys. Res. 87, 12311238.
Orsi, G., Gallo, G., Zanchi, A., 1991. Simple-shearing block resurgence in caldera
depressions.A model from Pantelleriaand Ischia.J. Volcanol. Geotherm. Res.47, 111.Orsi, G., Civetta, L., Del Gaudio, C., de Vita, S., Di Vito, M.A., Isaia, R., Petrazzuoli, S.M.,Ricciardi,G., Ricco, C., 1999a. Short-termground deformations and seismicity in thenested Campi Flegrei caldera (Italy): an example of active block-resurgence in adensely populated area. J. Volcanol. Geotherm. Res. 91, 415451.
Orsi, G., Petrazzuoli, S., Wohletz, K., 1999b. Mechanical and thermo- uid behaviourduring unrest episode at the Campi Flegrei caldera (Italy). J. Volcanol. Geotherm.Res. 91, 453470.
Orsi, G., D'Antonio, M., de Vita, S., Gallo, G., 1992. The Neapolitan Yellow Tuff, a large-magnitude trachytic phreatoplinian eruption: eruptive dynamics, magma with-drawal and caldera collapse. J. Volcanol. Geotherm. Res. 53, 275287.
Orsi, G.,de Vita,S., DiVito,M., 1996.The restless,resurgentCampi Flegreinested caldera(Italy): constraints on its evolution and conguration. J. Volcanol. Geotherm. Res.74, 179214.
Orsi, G., de Vita, S., Di Vito, M., Isaia, R., Nave, R., Heiken, G., 2003. Facing volcanic andrelated hazards in theNeapolitan area. In:Heiken, G.,Fakundiny, R.,Sutter, J. (Eds.),Earth Sciences in the Cities: A Reader: AGU Sp. Publ. Series 56, Washington, D. C.,pp. 121170.
Orsi, G., Di Vito, M.A., Isaia, R., 2004. Volcanic hazard assessment at the restless Campi
Flegrei caldera. Bull. Volcanol. 66, 514530.Pyle, D.M., 1989. The thickness, volume and grainsize of tephra fall deposits. Bull.
Volcanol. 51, 115.Pyle, D.M., 1995.Assessment of theminimumvolume of tephrafalldeposits.J. Volcanol.
Geotherm. Res. 69, 379382.Pyle, D.M., 2000. Sizes of volcanic eruptions. In: Sigurdsson, H., Houghton, B.F., McNutt,
S.R., Rymer,H., Stix, J. (Eds.), Encyclopedia ofVolcanoes. Academic Press,San Diego,pp. 263269.
Ricco,C., Aquino,I., Borgstrom, S.E., DelGaudio, C.,2007. A study of tilt change recordedfrom July to October 2006 atthe Phlegraean Fields (Naples, Italy). Ann. Geophys. 50(5), 661674.
Romano, C., Giordano, D., Papale, P., Mincione, V., Dingwell, D.B., Rosi, M., 2003. The dryand hydrous viscosities of alkaline melts from Vesuvius and Phlegrean Fields.Chem. Geol. 202, 2338.
Selva J., Costa, A., Marzocchi, W., Sandri, L., submitted for publication-a. BET VH: Long-term hazard from tephra fallout at Campi Flegrei, Italy, Bull. Volcanol.
Selva, J., Orsi, G., Di Vito, M., Marzocchi, W., Sandri L., submitted for publication-b.Probability hazard map for future vent opening at the Campi Flegrei caldera, Italy.
J. Geophys. Res.
Sigurdsson, H., Houghton, B.F., McNutt, S.R., Rymer, H., Stix, J., 2000. Encyclopedia ofvolcanoes. Academic Press, San Diego.Simkin, T., Siebert, L., 1994. Volcanoes of the world: a regional directory, gazetteer, and
chronology of volcanism during the last 10, 000 years (second edition).Smithsonian Institution and Geoscience Press Inc., Tucson.
Tilling, R.I., 1989a. Volcanic hazards and their mitigation: progress and problems. Rev.Geophys. 27, 237269.
Tilling, R.I. (Ed.), 1989b. Volcanic Hazards: Short Course in Geology 1. InAGU,Washington D.C.. A Chinese-language version is available in:.Tilling, R.I., 1990b.
J. Seismology (Beijing, China) 3, 2 (39), 146 A Spanish-language version is avail-able in:.Tilling, R.I., 1993b. Los Peligros Volcnicos. World Organization of VolcanoObservatories (WOVO-IAVCEI). (Translation by Ing. Bernardo Beate, Ecuador).
Tilling, R.I., 2005. Volcano hazards. In: Marti, J., Ernst, G.G.J. (Eds.), Volcanoes and theEnvironment. Cambridge University Press, Cambridge, pp. 5589.
Tonarini, S., D'Antonio, M., Di Vito,M.A., Orsi,G., Carandente, A., 2009. Geochemical andB-SrNd isotopic evidence for mingling and mixing processes in the magmaticsystem that fed the Astroni volcano (4.13.8 ka) within the Campi Flegrei caldera(Southern Italy). Lithos 107, 135151.
Walker, G.P.L., 1973. Explosive volcanic eruptions a new classication scheme. Geol.
Rundsch. 62, 431446.Walker, G.P.L., 1980. The Taupo Pumice: product of the most powerful known
(ultraplinian) eruption. J. Volcanol. Geotherm. Res. 8, 6994.Woo, G., 2008. Probabilistic criteria for volcano evacuation decision. Nat. Haz. 45,
8797.Zollo, A., Maercklin, N., Vassallo, M., Dello Iacono, D., Virieux, J., Gasparini, P., 2008.
Seismic reections reveal a massive melt layer under Campi Flegrei volcanic eld.Geophys. Res. Lett. 35, L12306.
276 G. Orsi et al. / Earth and Planetary Science Letters 287 (2009) 265276