Jökull - 01.12.1989, Blaðsíða 16

flows. Direct evidence from observations since
1919 as well as historical records of jökulhlaups,
extending back to the sixteenth century, suggests
that the caldera lake has existed for at least 400
years. At the same time numerous eruptions are
known to have occurred in Grímsvötn (Þórarinsson,
1974). The lavas must therefore have been extruded
subaqueously. The edges of the lava flows are 15-
20 m high and their surface is relatively flat apart
from a slight dip towards north.
The lava flows observed on the present caldera
floor were formed in eruptions in the southern part
of the caldera. The dip of the deeper reflections is
similiar to the general dip of the caldera floor. This
suggests that the lavas, that are believed to give rise
to these reflections, also originated in the southem
part. Apparently, eruptions have been more frequent
under the southem caldera wall than elsewhere
within the caldera.
Most of the lava flows are believed to be small.
Each flow covers an area of only a few square
kilometres. This implies that the volume of each
flow is probably less than 0.1 km3. Larger flows
would cross the caldera floor to the northem caldera
wall. The deeper reflections in line 2 do indeed sug-
gest that larger lava flows have been erupted occa-
sionally. Altematively, these deeper units could be
sills, intruded into the caldera infill. Further, the
data suggest that the caldera infill is mostly made up
of lavas in the southem and southwestern part of the
caldera but that sediments constitute a greater pro-
portion in the northern part (Fig. 8).
The contribution of volcanic eruptions within the
caldera to the melting of ice has been considered
negligible, as the eruptions have been considered
small, producing mostly tephra (Þórarinsson, 1974).
The existence of lava flows within the caldera sug-
gests, that these ideas need to be reconsidered. The
flows could be produced by the eruptions that have
been observed at the end of some jökulhlaups. They
could also be produced by entirely subaqueous erup-
tions which were not detected, probably occurring at
high water level. If eruptions producing lava flows
have been frequent, their effect on the thermal heat
flux of the area could be substantial.
Bjömsson (1983, 1988) estimated variations in
subglacial melting at Grímsvötn over the period
1860-1986. These variations reflect variations in the
thermal power of the area. His results show a base
heat transfer rate of 4500-5500 MW. Peaks in the
heat flux when the total power reaches 10000-
15000 MW, are superimposed on the base heat flow.
Björnsson suggests that these peaks are caused by
volcanic eruptions at the glacier base, but that the
base heat flux is drawn from a deeper magma body
by hydrothermal convection. A decline in the base
rate from 5000 MW to 4500 MW is detected over
the period up to 1976. Since 1976, a sudden drop in
the thermal power to 2500-3000 MW has been
observed. This decreased geothermal output of
Grímsvötn is further supported by the measurements
mentioned earlier, on the increased thickness of the
ice shelf. Also, maps compiled at various times over
the last 50 years show that the area of the lake has
decreased in this period (Bjömsson, 1988).
As pointed out by Björnsson (1974, 1988), an
eruption within the lake is unlikely to spark off a
jökulhlaup immediately, as the water level would be
raised only slightly. This can be explained by the
fact that melting of the ice shelf does not change the
water level, and rising of the water level would be
controlled by the flow of ice into the lake. On the
other hand, a subglacial eruption north of the lake
could melt large volumes of ice and the meltwater
would be drained into the lake causing a sudden rise
in water level. The jökulhlaup in 1938 was an
example of such an event (Björnsson, 1988). There-
fore, a difference exists between the effects of vol-
canic eruptions on the behaviour of the jökulhlaups:
frequent eruptions within the caldera are likely to
sustain a high level of melting but raising of the
water level would be gradual. On the other hand, an
eruption north of the caldera would most likely
spark off a large unexpected jökulhlaup. Conse-
quently, a sharp peak would be detected in the
observed thermal power of the area.
The thermal effects of eruptions onto the lakefloor
can be estimated in the following way: Consider a
period of high eruption frequency, with an eruption
occurring once every 10 years. In each event, a total
14 JÖKULL, No. 39, 1989