Page 325 - Caldera Volcanism Analysis, Modelling and Response

300 Valerio Acocella

with one set of reverse and normal faults; higher ratios (type B; Section 3.2) are
associated with multiple sets of reverse faults and eventually, by a set of normal faults
on top (Figure 4). However, this discrepancy is purely apparent. In fact, in type B
experiments, multiple reverse faults are required to propagate the collapse upwards
in a thicker crust analogue. Normal faults, as resulting from the gravitational
collapse of the wedges above the upper reverse ring fault at surface, form only if the
displacement along this upper fault reaches a certain threshold. This is exactly what
is observed with type A experiments. Therefore, independently from the roof
aspect ratio, all the collapses may display one or more set of reverse faults; if the
displacement on the upper ring fault at surface reaches a certain threshold, a normal
ring fault may also form.
In synthesis, despite minor variations in the amount and location of the ring
faults, the evolution of all the experimental collapses is characterised by outward
dipping reverse ring faults and, after a certain amount of slip, inward dipping
normal ring faults at the periphery; both ring faults replace former inward tilts. This
behaviour is observed with different apparatus and materials (sand, flour or clay as
brittle crust analogue and air, water and silicone as magma analogue), scaling (times,
strain rates and lengths), topography (with or without volcanic edifices, with
various slope dips), stress conditions (neutral, compressional, extensional) and
caldera elongation. Interestingly, the same structures were also obtained in previous
experiments investigating differential uplift (Sanford, 1959) and the depletion of
reservoirs (Odonne et al., 1999). Such a general consistency indicates that the
overall deformation pattern during collapse is, in the experiments, independent
from the strength of the used brittle crust analogues, the viscosity of the magma
analogue, the duration and size of the experiment and the presence of a regional
stress field or the load of any edifice. Moreover, this consistency indicates a precise
and constant structural behaviour in accommodating the room problem during
collapse, inferring a wide applicability of the analogue results.
Despite this general agreement among the various experimental data, some
minor discrepancies do exist. Possibly, the most relevant regards the development
of trapdoor collapses. Trapdoor collapses were common in Roche et al. (2000),
Walter and Troll (2001) and Kennedy et al. (2004), and rare in Acocella et al.
(2000). Some of the obtained trapdoor structures were expectable from the
imposed conditions, as the shape of the chamber analogue (Acocella et al., 2001)or
the distribution of the roof load (Kennedy et al., 2004). Other trapdoors were
unexpected, and may have occurred as a result of heterogeneities in the system,
often difficult to control a priori (Roche et al., 2000). Moreover, in Acocella et al.
(2000), the most subsided part of the trapdoor is always located above the top of the
asymmetric domed reservoir. Conversely, in Kennedy et al. (2004), the most
subsided part of the trapdoor collapse is above the most depressed area of the
reservoir. This apparent discrepancy can be explained by the fact that, when the
subsidence is homogeneous, the faults form along the points of maximum curvature
of the top of the reservoir; these are usually located in the most uplifted part of the
reservoir. As a consequence, the maximum subsidence is observed above the most
uplifted part of the reservoir (Acocella et al., 2001). Conversely, when the
subsidence within the reservoir is differential (as when deflating a balloon), the most