Jökull - 01.06.2000, Qupperneq 71

Páll Einarsson and Bryndís Brandsdóttir
Let us, for the sake of argument, assume that all
the melt water finds its way into the crust, and that
the porosity of the crust is 10%. This would raise the
groundwater level and increase the pore pressure by
)  $&2 3' 2  $12 546,.0' $1%879$12;: <,/.0'( $1%
( $12 is density of water, ' 2 increase in groundwater
level.)
The groundwater level is, thus, raised roughly
nine times the thickness reduction h of the glacier,
which leads to pore pressure increase that is ten times
larger than the load term subtracted from the principal
stresses. If the porosity is smaller, this amplification
factor becomes even greater. It is clear that even if
only a small fraction of the melt water (of the order
of the porosity) goes into the crust it will lead to pore
pressure changes that are significant compared to the
load reduction.
The triggering process described above should be
effective in seismically active areas wherever there
is significant accumulation of snow which is subse-
quently melted. The presence of the glacier is not
required. Why then is the annual cyclicity only ob-
served beneath the Mýrdalsjökull glacier and not in
other comparable areas, such as Torfajökull, Hengill,
Bárðarbunga or along the Loki Ridge? As noted
above, the triggering effect only works in a finely
tuned mechanical system. The rate of stress loading
must be comparable to the rate of stress modula-
tion by the trigger. If the stress loading is faster the
seismicity will be continuous, possibly only slightly
modulated by the trigger. If the stress loading is very
slow, the seismicity will be low and the trigger will
have little effect. It is, therefore, not to be expected
that all seismic areas with seasonal snow load will
show annual seismicity. Furthermore, the larger the
amplitude of the trigger the more likely it is to have
a noticeable effect. We note that the Mýrdalsjökull
area is the area of the highest snow accumulation
in Iceland. Annual accumulation of 3-9 m of snow
is reported (Magnús Tumi Guðmundsson, personal
communication, 2000). The annual amplitude of the
loading-deloading effect is, therefore, of the order of
0.003 MPa and the pore pressure effect of the order of
0.03 MPa.
The main conclusions of this paper may be summa-
rized as follows:
1. The seismicity of the Mýrdalsjökull region
originates in two well separated clusters with
an area of 30-35 km

and 70-80 km

.
2. The depth of hypocenters cannot be well re-
solved, but all available data are consistent with
a shallow source, at 0-5 km depth.
3. The eastern epicentral cluster is within the
Katla caldera and coincides with the area of P-
wave delays and S-wave shadows thought to re-
flect a magma chamber at shallow levels in the
crust.
4. The western cluster lies west of the caldera and
is interpreted as a manifestation of a separate
volcanic center, here called the Goðabunga vol-
cano.
5. The earthquakes in the Mýrdalsjökull area have
a low-frequency characteristic, typical of many
volcanic areas. P-waves are emergent and S-
waves are frequently missing. These charac-
teristics are stronger for the Goðabunga earth-
quakes than the events in Katla.
6. The Mýrdalsjökull seismicity has a definite an-
nual cycle, with earthquakes preferentially oc-
curring during the autumn months. The annual
correlation is particularly strong for the Goða-
bunga epicentral cluster. This phenomenon
may be explained by the combined triggering
effects of reduced ice load after the summer’s
melting and elevated pore fluid pressure in the
underlying crust. A continuous stress loading
process is necessary to maintain the seismic-
ity, here suggested to be the strain due to plate
movements.
In spite of the explanations suggested in this paper,
the persistent seismicity of Mýrdalsjökull and its sea-
sonal variability continue to be enigmatic. The effects
of glacial loading and pore pressure fluctuations need
to be quantified further, both theoretically and, if pos-
sible, by observation.
70 JÖKULL No. 49