Jökull - 01.06.2000, Síða 22

Holocene eruptions within the Katla volcanic system,
lagoons, raise the sandur surface and enlarge the sand-
ur plain at the expense of vegetated areas, shift river
courses and extend the shoreline southwards.
DISCUSSION
The recurrence time of eruptions in each of the three
categories of volcanic activity within the Katla system
apparently differs by orders of magnitude. The intra-
caldera basaltic eruptions occur at intervals of decades
while the silicic eruptions occurred at intervals of cen-
turies and the "fires" involving both the central vol-
cano and its fissure swarm occur at intervals of thou-
sands of years. The past eleven centuries are atypical,
firstly because of the major "fires" in the 10th century,
and secondly because no silicic eruptions have been
identified during this period.
The eruption frequency presented in this paper
should be regarded as a minimum for the intracaldera
basaltic eruptions and silicic eruptions. Only sub-
glacial eruptions that broke through the ice cover and
left a tephra layer are recorded. Small subglacial
events such as the 1955 and 1999 events, some of
which may have been small eruptions, went unno-
ticed until this century. The long repose after the
Eldgjá eruption is defined by the absence of tephra
layers in soils formed during the following centuries,
and events that did not leave such evidence may have
taken place during this interval. Infrequent rifting
episodes resulting in major "fires" are, however, in ac-
cordance with the location of the system within a zone
of propagating rift.
The absence of silicic eruptions after the Eldgjá
eruption may imply that the magma system below the
central volcano was disrupted or reorganized as a re-
sult of that event. The fairly homogenous glass com-
position of the 12 tephras erupted in the interval be-
tween the two "fires" indicates that stable conditions
prevailed during that period. The fact that the com-
position of the Holocene silicics differs significantly
from the pre-Holocene silicics may similarily indicate
significant changes of subcaldera conditions at an ear-
lier stage, either as a result of the 10300 14C yrs BP
eruption (Lacasse et al., 1995; Bard et al., 1994) or of
the Hólmsá fires some 6800 14C yrs ago.
The hyaloclastite flow deposits at Kriki may be
the first Holocene deposits of this kind that have been
identified in Iceland. The relationship to the other
Eldgjá products is still not sufficiently clear and fur-
ther field work is needed to establish an unambigu-
ous correlation. The apparent low productivity of air-
borne tephra within the caldera is, however, readily
explained if most of the magma erupted there escaped
from the caldera as a hyaloclastite flow.
Changes in subglacial drainage routes of post-
Eldgjá jökulhlaups towards the Kötlujökull gap can
have several causes. Different ice thickness within
the caldera as a result of different climate conditions
cannot be disregarded, e.g. migration of the ice di-
vide, which Dugmore and Sugden (1991) proposed to
explain changes in maximum advances of Sólheima-
jökull during the Holocene. Other plausible causes
for changed drainage routes are a shift in location of
intra-caldera eruption sites or altered caldera topogra-
phy. The latter would also affect the outlet glaciers
and might explain some of the changes observed by
Dugmore (1987) and others. Considering the magni-
tude of the Eldgjá event and the activity within the
caldera, changes of caldera topography cannot be dis-
counted as a contributing factor, while an eastward
shift of post-Eldgjá eruption sites is the most straight-
forward explanation.
Old traditions regarding settlements in the
Mýrdalssandur area become understandable in light
of the changes caused by the Eldgjá eruption. Large
parts of the area now known as Mýrdalssandur were,
without doubt, favourable for farming before the Eld-
gjá eruption. In the two quiet centuries following
the eruption, some of the areas laid waste by lava
flows and jökulhlaups recovered, if only temporarily.
In such a case the existence of farms along the now
sand-covered western edge of the lava flow, as well as
elsewhere along the edges and on islands within the
lava, is readily explained. Tales of clusters (hverfi) of
farms, both at the time of the Norse settlement and in
the centuries following the Eldgjá eruption, in areas
now laid waste, do not contradict what is currently
known about the geological and geographical condi-
tions. These tales may therefore contain some large
grains of truth - or simply be true.
The greatest hazard in historical Katla eruptions
JÖKULL No. 49 21