Page 298 - Caldera Volcanism Analysis, Modelling and Response

A Review on Collapse Caldera Modelling 273

may have been active during previous eruptions, including the formation of the
caldera itself, or whether this melt has been recently replenished from greater depth
through deep-rooted plumbing systems. As a general picture, we can infer the
existence of multiple reservoirs situated at different depths beneath active calderas: a
deep lower crust/upper mantle magma reservoir is tapped by a plumbing system
which channels melt upwards to form mid-crustal magma reservoirs. Most of these
reservoirs appear to have flat sill-like tops with diameters matching the diameter of
the superficial collapse structure. What is important to note is that low-velocity zones
overlying these mid-crustal reservoirs indicate the coupling effects of magma reservoirs
into a third level of reservoirs, shallow-seated hydrothermal systems. These coupling
effects have recently been inferred as the predominant source of recent unrest at some
of the investigated calderas here (Battaglia et al., 2006; Foulger et al., 2003; Gottsmann
et al., 2006a, 2006b; Todesco, 2008; Gottsmann and Battaglia, 2008).
The available information on caldera bounding faults and their geometry from
geophysical studies can be regarded as inconclusive at best. While fault planes of
active regional faults can be imaged rather accurately, for example, at Long Valley
(Prejean et al., 2002)(Figure 15), images of lateral discontinuities associated with
bounding faults are blurred, particularly at shallow levels. This could be due to two
factors: explosive calderas are systems with activities spanning over tens to hundreds
of thousands of years. Experimental work shows that resurgence over collapse
(and vice versa) is characterised by the complete reactivation, with opposite
kinematics, of all the pre-existing ring faults during inversion (Acocella et al.,
2000). Continual kinematic movement along bounding faults over the lifespan of
an explosive caldera is likely to initiate strong weakening of the bounding rocks as
well as the creation of secondary faults. Second, bounding faults are important
escape routes for magmatic fluids from the reservoir to the ground surface as
evidenced by intense hydrothermal activity, alteration and magnetic lows along the
walls at many active calderas. Prolonged corrosion of fault walls and the presence
of fluids within the fault system significantly affect the mechanical properties of the
surrounding rocks as well as their coupling to seismic waves.
The available data suggest that bounding faults are, averaged over their entire
imaged length, subvertical with a trend to a slight inward dip towards the caldera
centre. Assuming that these faults are representative of the faults active during the
collapse event, the data do not provide a solution to the space problem. However,
bearing in mind the complexities at shallow depths (as imaged, for example, at
Rabaul in Figure 12) and results of fault alignment in analogue models (at systems
of both inward- and outward-dipping faults), the same faults may be active during
the formation of a caldera as well as throughout its post-collapse evolution.

5. Discussion and Implications

5.1. On the use of experimental models
Experimental models on collapse calderas enable a qualitative study of the structural
evolution of a collapse process and suggest which factors play a more relevant role.