Jökull - 01.01.2005, Qupperneq 66

Bergrún Arna Óladóttir et al.
also shows that 99 layers have a Fe-Ti basalt compo-
sition typical for the Katla system and seven of those
layers contain sporadic grains of silicic glass. Two
layers are basaltic icelandites of probable Katla ori-
gin because of their high Fe-Ti content. Other dark-
coloured prehistoric tephra layers in the composite
profile belong to the Grímsvötn, Veiðivötn and Hekla
volcanic systems (Figure 5). The prehistoric part of
the composite section also contains 16 felsic layers,
of which ten are SILK layers. Only two of these lay-
ers were analysed in this study (i.e. SILK-N4 and -A7)
because most of them have been analysed previously
by Larsen et al., (2001). The remaining six felsic lay-
ers are the knownmarker tephra layers from the Hekla
and, most likely, the Torfajökull volcanic systems.
The Katla tephra time-series reveals distinct pe-
riodic patterns in major-element composition of the
basaltic tephra layers, illustrated by SiO2 concentra-
tions versus time (Figure 7). Three compositional
patterns are identified: (a) periods where incompati-
ble elements increase steadily with time, (b) periods
with no discernable change in the major element com-
positions (stable conditions), and (c) periods where
changes in magma composition appear to be random
with respect to time (Óladóttir, 2004). Each of these
patterns are thought to reflect distinct conditions for
magma supply and evolutionwithin the plumbing sys-
tem beneath the Katla volcano. The exact nature of
these compositional patterns and the implication they
may have for the plumbing system of the volcano is
beyond the scope of this paper and will be discussed
elsewhere (Óladóttir et al., in preparation).
However, on the basis of these temporal variations
in composition, we divide the Holocene activity into
eight periods, here labelled I-VIII (Figure 7). Six of
the periods are relatively short, or between 510-980
years long. Two are fairly long, each spanning about
1700 years (Figure 7 and Table 4). A cyclic behaviour
is observed where a constant value (period VIII) is
followed by an irregular one (VII) which is then re-
placed by a gradual change (VI). This cycle repeats
itself through periods V-II and the last period (his-
torical time) demonstrates the beginning of the third
cycle.
Eruption frequency of basalts
The proportion of Katla layers in the soil profile as
identified in the field is similar to that obtained using
chemical composition as the discriminator. On the ba-
sis of field identification, using criteria such as colour,
grain size and clast morphology, 172 out of 208 tephra
layers, or 83% of all measured layers, where assigned
to the Katla volcanic system. Of these 172, we mea-
sured the chemical composition of 111 layers (repre-
senting 88% of the 126 layers analysed). In all cases,
the assigned Katla origin was confirmed by the ob-
tained chemical composition. Therefore we are confi-
dent in our field identification and use the number of
Katla layers identified during field measurements in
our calculations of the Holocene eruption frequency.
The true eruption frequency at a given volcano
can only be obtained if all its eruptions are known.
This is most easily achieved for time periods where
reliable eyewitness or written records exist. For peri-
ods where such records are not available, such as the
Holocene, the accuracy of the calculated eruption fre-
quency depends on the quality of an acquired data set
or, in the case of this study, the completeness of the
tephra stratigraphy. In this context it is important to
recognize that tephra dispersal is primarily controlled
by the eruption intensity and wind direction and to
some degree the eruption duration, because the dis-
persal of tephra in sustained events is more strongly
affected by changes in eruption intensity and wind di-
rection. Consequently, a complete record of explosive
eruptions within a volcanic system such as Katla re-
quires detailed measurements of the tephra stratigra-
phy in soil profiles at proximal and medial locations
circumscribing all possible vent regions. This is, how-
ever, only possible in regions where the soil has been
conserved or at least where different soil sections are
not too far apart.
This study is confined to tephra layers deposited
east of the Katla volcano. Comparison of the number
of tephra layers found between marker tephra layers
in soil sections east and west of the volcano shows
a certain similarity. The similarity between eruption
frequency patterns to the east and west of the volcano
is sufficiently close to derive that of the volcano itself
from the eastern soil sections. During historical time,
66 JÖKULL No. 55