Page 293 - Caldera Volcanism Analysis, Modelling and Response

268 J. Martı ´ et al.

Their combined horizontal extent matches the long axis of the 30 100 km wide
caldera. The location of a pronounced gravity low within the collapse structure
coincides with one of these reservoirs. The combined results suggest the presence
of, as yet, the world’s largest magmatic reservoir at mid-crustal levels beneath Toba
(Masturyono et al., 2001).

Deep-seated reservoirs (or deep extension of magmatic bodies). In addition to
the two large mid-crustal reservoirs, the seismic image at Toba also reveals a low-
velocity column with a reduction in P-wave velocities similar to those of the mid-
crustal reservoirs which can be traced from one of the reservoirs into the uppermost
mantle (Masturyono et al., 2001). The extent of this low-velocity zone is matched
by the planar distribution of low-frequency earthquakes in the range of 20–40 km
depth. A picture of a deep root of magma replenishment also emerges from Valles,
where a large planar zone of reduced P-wave velocities ( 15%) can be mapped out
in the lower crust or upper mantle (Aprea et al., 2002). This zone is interpreted to
represent melt formation in the upper mantle and subsequent basaltic underplating.
At Long Valley, the tabular mid-crustal reservoir may represent the top of a diapir-
like ridge rising up from the migmatized lower crust of the Basin and Range
province (Steck and Prothero, 1994). At Las Can ˜adas, the magnetic anomaly
appears to be rooted in a dyke complex down to ca. 16 km bsl (Aran ˜a et al., 2000).

The nature of reservoirs. While shallow-level low-velocity zones are generally
regarded to represent hydrothermal reservoirs, deeper-seated anomalies are usually
associated with the presence of a melt phase. However, the upper crust around a
volcano is composed of rocks, which are expected to show large variations of
seismic wave velocity due to significant variations of composition, fluid content,
porosity, and temperature. As a consequence, ray paths are undoubtedly 3-D,
adding a high degree of non-linearity to the relationship of ray path and velocity.
Thus, assessing the melt fraction within a low-velocity zone is challenging and
potentially associated with large errors. Weiland et al. (1995) give a short discussion
on the ambiguities involved in estimating melt percentage in low-velocity zones
and compare results obtained from different models. For example, the tabular
30% low-velocity zone at Long Valley is regarded as containing a melt fraction
anywhere between 7 and 100%. At Valles, 100% melt is attributed to the central
ellipsoidal 30% velocity (Figure 13). Masturyono et al. (2001) infer melt-
dominated reservoirs with o10 km thickness in the mid-crust and a 0.1 melt
fraction in the deep root of the plumbing system.

4.1.2. Caldera bounding faults
Geophysical information on the geometry of bounding faults is scarce and stems
predominantly from seismic investigations. One of the key questions regarding
bounding faults and their role in caldera evolution concerns their dip angle as
well as their dip direction: vertical, inward, or outward dipping with respect to the
centre of the caldera.