Jökull - 01.12.2006, Side 62

David W. McGarvie et al.
One solution to this stratigraphic problem is to de-
termine the absolute ages of representative volcanic
units. However this poses two problems: firstly that
a dominance of subglacial eruptives may require a
large number of samples to be dated before even a
rudimentary stratigraphy can be established, and sec-
ondly that at the present time the dating of such young
rocks is technically difficult, and even the most well-
developed method (Ar-Ar dating) is limited because
of the high uncertainties generated when analysing
young rocks with low potassium concentrations (see
Hawkesworth et al., 2004, and references therein).
Rocks high in potassium are preferred because they
give both lower uncertainties and more meaningful
ages, but Iceland’s most K-rich volcanic rocks are
evolved rocks (mostly rhyolites but including tra-
chytes and mugearites) that are restricted to the flank
zones (Sæmundsson, 1974, 1979). With these con-
straints in mind, a realistic initial approach to unravel-
ling the morphological and geochemical evolution of
a flank zone central volcano is to date a small number
of carefully-chosen samples that form a skeletal but
robust stratigraphy constrained by absolute age deter-
minations.
The Torfajökull central volcano
The Torfajökull central volcano is Iceland’s largest
rhyolite complex (Walker, 1966) and lies in the south-
ern Iceland flank zone (Figure 1). Its volcanic prod-
ucts comprise c. 250 km3 of subglacial and subaerial
rhyolite (with minor amounts of basalt-intermediate
compositions) which cover an area of c. 450 km2
(Sæmundsson, 1972; McGarvie, 1984). The central
volcano is essentially a dissected rhyolite plateau
within which a suspected caldera measuring 18 km
by 13 km has been mapped by Sæmundsson and
Friðleifsson (2001) (Figure 2). Incomplete and poor
exposure of this elliptical structure does not allow
for the amount of downsagging or the nature of any
syn-caldera fill to be ascertained. Vigorous and per-
sistent geothermal activity is concentrated within the
suspected caldera, and this has led to extensive hy-
drothermal alteration of the rocks within and adjacent
to the structure (Friðleifsson and Sæmundsson, 2001).
Shallow seismicity focussed in central areas suggests
that shallow magma intrusion may be ongoing (Soos-
alu and Einarsson, 2004; Soosalu et al., 2006).
Torfajökull is still active, and its last eruption in
1477 AD produced two small rhyolite lava flows on
its northern flanks (Sæmundsson, 1972, 1988; Larsen,
1984). Volumetrically, Holocene rhyolites represent
only c. 1% of the exposed volcanic pile (about 2 km3),
and yet (compared to the Pleistocene rocks) they have
been exceptionally well studied (e.g. Blake, 1984;
Larsen, 1984; Mörk, 1984; McGarvie, 1984; Mac-
donald et al., 1990; Gunnarsson et al.; 1998; Stecher
et al., 1999), due in large part to their good exposure,
accessibility, and the existence of a relative stratigra-
phy (Sæmundsson, 1972, 1988; McGarvie, 1985) un-
derpinned by some tephrochronology (Larsen, 1984).
A comparable understanding of the pre-Holocene
geochemical and morphological evolution of Torfa-
jökull has been hampered by the lack of a sufficient
number of geochemical analyses from samples of un-
equivocal stratigraphic position within the volcanic
pile. Apart from information available in unpublished
theses (McGarvie, 1985; Ívarsson, 1992), geochemi-
cal analyses of Pleistocene rhyolites have been pub-
lished only in McGarvie (1984) and McGarvie et
al. (1990). These studies have emphasised an impor-
tant distinction between Torfajökull’s Holocene and
Pleistocene activity, with Holocene volcanism dom-
inated by less-evolved (i.e. subalkaline) and small-
volume (<0.3 km3) rhyolite eruptions accompanied
by modest-to-extensive interactions with mafic mag-
mas, whereas Pleistocene rhyolites are dominated by
more evolved (peralkaline) compositions involving
less extensive (or no) interactions with mafic magmas
(McGarvie et al., 1990). Additionally, available ev-
idence suggests that individual Pleistocene eruptions
were generally larger than the Holocene rhyolite erup-
tions (McGarvie, 1985; Ívarsson, 1992; Sæmundsson
and Friðleifsson, 2001), with the most extreme exam-
ple being the group of subglacial rhyolite edifices that
surround the main rhyolite plateau (termed the “ring
fracture rhyolites”) that McGarvie (1984) hypothe-
sised were produced during one large eruptive event
that took place during the last (Weichselian) glacial
period (see Figure 2 for locations).
60 JÖKULL No. 56