Page 419 - Caldera Volcanism Analysis, Modelling and Response

394 Micol Todesco

hydrothermal fluids on unrest processes and (4) describes some model applications
to the Phlegrean Fields caldera. Simultaneous modelling of different independent
parameters has proved to be a powerful tool for understanding caldera unrest. The
results highlight the importance of comprehensive conceptual models that incorporate
all the available geochemical and geophysical information, and they also stress the need
for high-quality, multi-parameter monitoring and modelling of volcanic activity.

1. Introduction

Many active calderas are densely populated and thus require effective
evaluation of volcanic hazards. However, the quantification of volcanic hazards in
active calderas is a difficult task. This is because a large variety of eruptive styles and
intensities are possible and in addition the opening of new vents can potentially
affect wide areas, but the number and location of these is uncertain. To add to this
complexity, non-eruptive unrest is typical of caldera volcanic systems (Newhall and
Dzurisin, 1988; Hon and Pallister, 1995; Cole et al., 2005). Unrest crises
commonly involve ground deformation, gravity changes and seismic activity, in
addition to changes in composition, temperature or discharge rate of hydrothermal
fluids, regardless of the eruptive or non-eruptive nature of the crisis. Yet eruptions
may occur at calderas without significant warning. Unrest episodes have been
recorded at Long Valley since 1980 (Sorey et al., 2003, and references therein), and
at Phlegrean Fields since 1969 (Troise et al., 2008, this volume) without major
consequences during the following twenty years. At Rabaul (New Guinea) an
important unrest phase in 1983–1985 (McKee et al., 1984, 1985; Mori et al., 1989)
was followed by a relatively quiet period, whereas limited warning preceded the
onset of the 1994 eruption (Smithsonian Institution-GVN, 1994). Similar pattern
of unrest may lead to very different eruptive and non-eruptive scenarios; therefore
the identification of possible precursors of eruptive activity is difficult. Unrest
phenomena may occur as magma stored at depth approaches eruptive conditions,
i.e. when the ascent or intrusion of magma at shallow crustal levels modifies the
local stress field, affects temperature gradients, and is accompanied by exsolution of
magmatic volatiles. These processes are known to trigger typical unrest phenomena
that we can monitor at the surface, such as seismicity, ground deformation or
changes in geochemical and other geophysical parameters. During a volcanic crisis
the mitigation measures may involve partial or total evacuation of the population,
especially if a large explosive eruption is expected. In these cases crisis management
decisions rely to some extent on monitoring parameters likely to signal the onset of
eruptive. In the case of calderas chances of a false alarm are very high. The
occurrence of non-eruptive unrest crises, observed at several calderas in the world,
implies that the relationship between the magmatic system and unrest phenomena
may not be straightforward and that unrest crises are not necessarily synchronous
with magma ascent and evolution. Discrimination between eruptive and non-
eruptive crises is often possible only right before the onset of an eruption, and
evacuation may require significant time, depending on the number of people and