Environmental Engineering Reference
In-Depth Information
with each event. Over time, with a net magma supply from depth exceeding the
time-averaged outputs, the chamber enlarges (e.g. Reid,
2008
). With increasing
volume, the wall rocks heat up and deform in a more viscous fashion such that
the chamber overpressure reduces. As magma storage increases, rather than the
overpressure triggering eruptions, the buoyancy of the magma (whose density is
less than that of the host rocks) becomes the driving force for eruption. The critical
condition for a super-eruption is then the thickness of the reservoir since it in
u-
ences the competition between magma supply and cooling and crystallization
(the latter process being enhanced in thinner, wider reservoirs).
This theoretical
treatment might explain the in
ection apparent
in the
magnitude
frequency relationship for global (terrestrial) volcanism between
events of
M
7 and
M
8. Taking the statistics at face value, eruptions of size
M
8
and above are rarer than extrapolation of the data for smaller eruptions would
suggest (
Figure 2.1
). Further support for the hypothesis that buoyancy drives
super-eruptions comes from sophisticated measurements of the density of silicic
melts at realistic magmatic pressures and temperatures (Malfait
et al
.,
2014
).
-
Figure 2.1 Frequency versus magnitude plot for volcanic eruptions based on
records for last 300 years for eruption magnitudes
M
2
-
6; for the last 2000 years
for
M
6
of the past 45 Ma (after Mason
et al
.,
2004
). The average return period for each magnitude class is labelled on the
plot. The only
M
9 eruption in this compilation is the 28-Ma-old Fish Canyon
Tuff (Colorado, USA). The available data are imperfect and it is probable that not
all very great eruptions of the last few million years have yet been identi
-
8; and for all known
“
super-eruptions
”
ed.
Nevertheless,
the apparent fall off in the frequency of very great eruptions
(
M
8
-
ect a threshold between the contrasting driving
forces of eruptions (e.g. Caricchi
et al
.,
2014
).
9) may be real and re
