Page 265 - Caldera Volcanism Analysis, Modelling and Response

240 J. Martı ´ et al.

consequence of a pressure decrease within the chamber; (ii) simulating the
development of caldera collapse at the summit of a doming structure caused by a
pressure increase inside a magma chamber; (iii) simulating the effect of the
superposition of inflation (doming) and deflation episodes on the development of
collapse calderas; (iv) simulating the development of a collapse caldera, considering
the topographic loading of a volcanic edifice; and finally (v) those simulating
pure collapse due to magma chamber decompression, considering the role of pre-
existing structures and the effect of superimposing a regional tectonic regime
(deviatoric stress) on the local stress field.
The most common case reproduced experimentally corresponds to the pure
caldera collapse (i.e. subsidence takes place as a consequence of a pressure decrease
within the chamber without previous inflation) (Komuro, 1987; Martı ´ et al., 1994;
Roche et al., 2000; Acocella et al., 2000, 2001; Roche and Druitt, 2001; Walter
and Troll, 2001; Kennedy et al., 2004; Lavalle ´e et al., 2004; Geyer et al., 2006).
In this case, experimental results show that this type of collapse entirely depends on
the roof aspect ratio R. For low values of R (i.e. when Ro0.7–0.85) the caldera
collapse is piston-like, whereas for higher roof aspect ratios (RW0.7–0.85) the
collapse tends to be of incoherent or funnel type (Geyer et al., 2006). In the first
case, the subsiding block is coherent and collapse is controlled by a combination of
outward dipping reverse faults and vertical or subvertical ring faults (Figure 3A). In
the second case, collapse occurs along multiple reverse faults. At first, a pair of faults
nucleates at the top of the magma chamber and propagates upwards triggering
the formation of a second pair, and so on until the structures arrive at surface
(Figure 3B). No peripheral faults are generated, and subsidence is greater at depth
than at the surface. Collapse begins in both cases with downward flexuring of the
roof (‘down sagging’); however, this effect is more marked for low values of R.
Also, the percentage of extensional area in relation to the total collapse area
increases with R.
We can also observe in these experiments that the resulting caldera is characte-
rised by two depressions (Figure 3). The inner one is bordered by outward-dipping,

Figure 3 Photographs and sketches of two of the experiments conducted by Roche et al. (2000).
(A) R ¼ 0.2 and (B) R ¼ 2. R, reverse fault; N, normal fault (modi¢ed after Roche et al., 2000).