Page 329 - Caldera Volcanism Analysis, Modelling and Response

304 Valerio Acocella

Rabaul (Papua New Guinea) (Newhall and Dzurisin, 1988; Hallinan, 1993;
Skilling, 1993; Acocella et al., 2001, and references therein; Acocella et al., 2002;
Mueller and Mortensen, 2002). In particular, the Miyakejima collapse in 2000 is
possibly the most significant caldera-collapse episode (with subsidence of several
hundreds of m and fracturing) observed so far: the authors clearly describe the
development of an inner outward dipping reverse ring fault, formed first, and a
subsequent outer inward dipping normal fault, with an architecture and evolution
identical to that in the experiments (Figure 11f; Geshi et al., 2002).
The experimental results suggest that these concentric pairs of nested calderas
may be related to the same eruptive episode (Marti et al., 1994; Acocella et al.,
2001), not necessarily to two distinct eruptions, as often believed (e.g. Newhall and
Dzurisin, 1988, and references therein). In these cases, one can only discriminate
between the two possibilities constraining the relative timing of development of the
two collapses, considering not only surface data. In fact, at natural nested calderas,
surface data can only indicate that the outer caldera is older, as otherwise the inner
one would have been covered by the activity and deposits of the outer. For
example, surface data suggest that several pairs of nested calderas are characterised
by an inner younger structure, as Campi Flegrei, Latera, Tavua (Orsi et al., 1996;
Capaccioni et al., 1987; Setterfield et al., 1991). Interestingly, available subsurface
data at these same calderas indicate that the inner depression coincides with an older
buried collapse, later reactivated. This is shown, for instance, by the deep section of
Campi Flegrei, obtained from borehole and geophysical data (Figure 11; modified
after Orsi et al., 1996); the dips of the faults are approximate. The different
thickness of the syn-collapse deposits within the two nested structures suggests that
the inner, deeper depression formed before the outer one. Moreover, the presence
of syn-collapse deposits in both depressions suggests that there has been interaction
between the depressions during the same collapse event. These features are in close
agreement with the evolution and timing of formation of the experimental
collapses. These examples suggest that surface geology alone is not sufficient to
evaluate the relative timing of development of the depressions and, when such an
age is inferred only from surface data, it is not sufficient to define whether a pair of
nested calderas is consistent with Stage 4 or results from two distinct collapses.
Several large calderas in nature, characterised by a low aspect ratio of the
chamber roof (t/do0.5) and significant subsidence (W2,000 m), have only one
major ring fault visible at their borders. Notable examples include Long Valley
(Carle, 1988), Valles (Self et al., 1986), La Garita and Creede (Lipman, 1997, 2003),
Western United States and Okueyama ( Japan; Aramaki et al., 1977). In the light of
the experiments, these calderas, characterised by an advanced subsidence, can be
interpreted as being bordered by two ring faults lying next to each other. Their
proximity makes it difficult to resolve each structure, resembling, at the caldera
scale, a single ring system. A very similar deformation pattern, with almost
coincident ring faults, is shown in the experiments with t/do0.5 and significant
subsidence (Figure 4b; Roche et al., 2000).
Asymmetries in the development of these structures may result in trapdoor
collapses, commonly observed in the experiments and in nature, as at Grizzly Peak
(Colorado; Fridrich et al., 1991), Bolsena, Latera (Italy; Nappi et al., 1991) and