Page 321 - Caldera Volcanism Analysis, Modelling and Response

296 Valerio Acocella

These nucleate from the centre of the reservoir towards its periphery, as a result of
the progressive enlargement of the evacuated area at the top of the dome-shaped
bladder. This results in an outward incremental caldera growth (Figure 7b).
Subsequently, a peripheral outer set of inward-dipping normal faults develops.
Increasing the depth to the bladder decreases the width of the collapsed area, its
symmetry and coherence. Despite the overall similarity and consistency of the
deformation pattern with regard to previous models, minor differences exist. These
mainly concern the outward incremental growth of the caldera, conversely to what
observed in experiments with flat roof, where the width of the caldera is constant
from the beginning (Figure 7c; Acocella et al., 2000; Roche et al., 2000). Also, here
both the inner and outer set of faults show a polygonal shape (in map view), very
seldom previously observed. Finally, conversely to Acocella et al. (2001), trapdoor
subsidence is larger where the reservoir is deeper; this results from the larger
amount of local subsidence within the reservoir.

3.3. Caldera collapse under a regional stress field

Acocella et al. (2004) use the same apparatus and materials as Acocella et al. (2000).
However, here the sand-pack undergoes ‘‘regional extension’’ (achieved by the
lateral sliding of a basal sheet) only before the collapse starts. As the strain rates
10 1
during caldera collapse (eB10 s ) are much faster than those induced by
15 1
regional tectonics (eB10 s ), regional extension during collapse may be
neglected in the models. Therefore, these experiments simulate only the effect of
pre-existing regional normal faults on collapse, not of a coeval regional stress field.
However, they are useful to recognise the effect of the existing regional stress field
in controlling the geometry of the magma chamber at depth. The regional faults are
subparallel, mainly with similar dip and plunge (inward dipping atB601), bordering
a graben or half-graben structure. The main effect of the pre-existing structures on
collapse is their partial reactivation during the development of the outward dipping
reverse faults. This occurs only when the reverse fault is nearby and subparallel to
any regional normal fault (Figure 8). The reactivation forms a wider caldera along
the direction perpendicular to the normal faults, resulting in an elliptical depression
at surface. Its major axis is therefore parallel to the former extension direction. The
reactivation may account for the formation of elliptic calderas, with eccentricity
W0.8, elongated perpendicular to the rift axis. Moreover, this reactivation may
suggest an explanation for the presence of subsidiary depressions around the borders
of the calderas, influenced by both caldera collapse and regional tectonics (e.g.,
Spinks et al., 2005).
Holohan et al. (2005) use the same apparatus as Walter and Troll (2001).
However, the brittle material undergoes regional extension or contraction
(achieved by the sliding of a side wall) during collapse (Figure 9a). Therefore,
these models, conversely to Acocella et al. (2004), simulate any possible effect of a
regional stress field coeval to collapse. The main result is the variation in the dip
angle of the outward dipping reverse faults bordering the caldera (Figure 9). In fact,
under regional extension, along a section parallel to the maximum extension,
the reverse faults are subvertical, rather than dipping at B701 (Figure 9c).