Volcanoes - Extreme Events: A Physical Reconstruction and Risk Assessment - page 214

eruptions. In the latter case, ash is often projected into the upper troposphere
and as a consequence can circle the globe. Comparisons between identified
tephra layers and possible source volcanoes are made by comparing the con-
centrations of major oxides within the layer. Tephra shards are often used for
this type of analysis. However, locating the source volcano by the geochemical
signature of these shards is not necessarily straightforward and extensive dating
of the stratigraphic sequence is often first required in order to gain an idea of
which volcanoes around the globe may have erupted at that time. Even then,
problems can occur as dating techniques have an inherent level of inaccuracy. For
example, a tephra layer marking a major volcanic eruption in AD 1459 occurs in
ice cores in Antarctica and Greenland. It was probably one of the largest volcanic
eruptions globally over the past 700 years. The eruption most likely occurred in
Vanuatu when the volcano Kuwae destroyed an island (Australian Antarctic Divi-
sion, 2004). Historical records and tree rings suggest that this eruption occurred
in AD 1453, six years earlier than that suggested by the Antarctic ice cores. But
even when this is taken into account, along with potential errors in dating the
ice core, the earliest date for this eruption (based upon the ice core record) is
AD 1456. Such potential differences in determining the age of a specific eruption
event will become even more exaggerated when using dating techniques such as
radiocarbon or luminescence which have considerably larger uncertainty mar-
gins than tree-ring and ice-core dating. And this is especially so where possible
eruptions are closely spaced in time. Another potential problem encountered
during the correlation of tephra layers with possible eruption sources lies in
theanalytical methods applied to the micro-analysis of tephra shards. It is often
difficult to compare data generated by different operators on different analytical
machines which can result in reduced robustness of the data analysis.
Charman and Grattan (1999)suggest that it is possible to overcome these
difficulties by using discriminant function analysis (DFA). This type of analysis
uses a classification model based on a reference data set containing the major
oxides from known tephras. The classification model consists of a series of dis-
criminate functions that can be plotted graphically on discriminate function
axes and applied to the unknown tephra. The unknown tephra is then classified
into one of the known groups of tephra and is given a known probability for
misclassification.
Tephras can also be identified by their magnetic characteristics, a technique
known as magnetostratigraphy. The type and concentration of magnetic miner-
als in tephra layers depends on the chemical composition of the tephra. Differ-
ent types of glass-encased magnetite grains and ferrimagnetic components can
be identified by their magnetic components and magnetic susceptibility. These
magnetic characteristics reflect the chemical composition of the source material