Page 297 - Caldera Volcanism Analysis, Modelling and Response

272 J. Martı ´ et al.

Second, there are a number of inherent ambiguities in interpreting geophysical
data:
(i) Uncertainties in physicochemical properties of subsurface materials (e.g.
temperature, pressure, chemical composition, porosity, permeability, elastic
moduli, structural relaxation times, viscosity, density, fluid content, and
composition) have a strong effect on the emerging subsurface image. For
example, seismic wave speeds are strongly dependent on viscosity and
temperature, whereas the identification of density anomalies is biased on the
assumed density contrasts at depth. Physico-chemical property values of melts
are more commonly determined (see Spera, 2000 for a recent review), and
they should be incorporated in the analysis.
(ii) Parameterisation, for example, of the regional velocity model employed during
seismic investigation is crucial to the interpretation of velocity contrasts.
(iii) Computational restrictions play an important role for multi-parameter
inversion when comparing modern results to those obtained decades ago as
increased CPU power has enabled more elaborate modelling techniques.
(iv) Modelling and mathematical frameworks pronouncedly bias geophysical
results, for example, when comparing forward modelling results to inversion
results from gravimetric or magnetic investigations. The same applies to
differences in results between 1.5-D, 2-D, or 3-D modelling. When evaluating
compressive wave speed decrease, the assumed shape melt is contained in
strongly affects the inferred melt fraction (Weiland et al., 1995).
(v) The non-uniqueness of results is one of the major limitations for geophysical
investigations to provide realistic images of the subsurface, and this remains
a major challenge. For example, results from gravimetric inversions are
ambiguous due to the fact that different density distributions at depth can cause
the same gravitational perturbation at the surface.
In summary, geophysical imaging reveals a complex arrangement of subsurface
reservoirs beneath active calderas. At shallow level (surface to a few kilometres depth),
hydrothermal reservoirs and altered rock appear to cause a reduction in wave
velocities as well as negative magnetic anomalies. Caldera bounding faults also
represent zones of increased hydrothermal activity and must be regarded as important
pathways for fluid flow from depth (Todesco, 2008; Gottsmann and Battaglia, 2008).
Magma reservoirs can undoubtedly be associated with low velocities at mid-crustal
depth. However, the question still remaining concerns the melt fraction in these
anomalous bodies. Current estimates from P-wave contrasts or v p /v s ratios in these
zones can be regarded merely as rough proxies and are less suitable to assess the
eruptability of a large subcaldera magma reservoir. Moreover, due to the resolution
limit of geophysical images, smaller-sized magma pockets are very difficult to
recognise, if at all. Assuming that small (monogenetic) eruptions may be fed by the
interaction of fluid-filled cracks (Takada, 1989, 1994), information from geophysical
images are of only limited value for the assessment of hazards during periods of
increased unrest. In assessing the melt fraction in large anomalous bodies, there is a
high degree of uncertainty regarding their relative ‘age’. It is obviously impossible to
assess whether the melt phase is a remnant of a crystallising magma reservoir that