Jökull - 01.01.2021, Síða 63

Bedrock and tephra layer topography within the Katla caldera
also brought the tephra layer down to the glacier bed,
beneath many cauldrons, revealed as data gaps in Fig-
ure 8a. It is not always clear what causes the absence
of the 1918 tephra layer in other areas but the glacier
in these areas is often relatively thin (e.g. areas near
Goðabunga at the south-west rim of the caldera and
survey areas outside the caldera).
In 2016, the 1918 tephra layer lay mostly at 200–
350 m depth, in few areas at less than 200 m depth, but
in the westernmost part of the survey area, the tephra
was at ∼100 m depth (Figure 8a). This shallow depth
of the layer, almost 100 years after the eruption, is
noteworthy given that the annual balance, measured at
the peak of Goðabunga ∼3 km further south, is typ-
ically between 3 and 4 m ice equivalent (Ágústsson
et al., 2013 and more recent unpublished data from
JÖRFÍ field surveys). A recent study applying snow
radar to map winter accumulation did however indi-
cate ∼40% less winter snow in the area of minimum
tephra layer depth in May 2016 than at the peak of
Goðabunga (Hannesdóttir, 2021). This is probably
a persistent accumulation pattern produced by redis-
tribution due to snow drift (e.g. Dadic et al., 2010)
reducing winter accumulation at a ridge between the
caldera plateau and the west side of Mýrdalsjökull,
explaining the shallow depth of the tephra at this lo-
cation (Hannesdóttir, 2021).
The topography of the 1918 tephra layer within
the ice is clearly shaped by subglacial geothermal ac-
tivity. A good example is the area at the caldera cen-
tre, where the subglacial geothermal activity, beneath
K5, K6, K7 and K19, clearly leave their imprints in
the 1918 tephra layer (Figure 11). We also observe
a strong depression in the tephra layer south of K19
(also seen as a dip in the traced tephra layer on Figure
2e). This (marked A in Figure 11) is not a location of
previously known cauldron, although there is a clear
curve in the surface contour lines matching the loca-
tion of the depression in tephra layer. Subglacial melt-
ing due to geothermal activity seems required to ex-
plain these features. Another less obvious feature is a
∼25 m deep depression in the tephra layer north of K6
(marked B in Figure 11). This depression is also ob-
served in the deeper tephra layer at this location (Fig-
ure 8b). At first glance, this shallow depression seems
likely to be formed by ice dynamics. However, com-
parison of the glacier surfaces from 2016 (Pléiades)
and 2010 (Jóhannesson et al., 2013) shows that minor
curves in the surface elevation contours at this loca-
tion in 2016 were substantially stronger in 2010; the
surface undulation at this location has been smoothed
from 2010 to 2016. This shallow depression in the
tephra layer is therefore likely an imprint of weak and
probably sporadic geothermal activity at the glacier
bed, which may have been dormant in 2010–2016.
The beautiful depression in the 1918 tephra layer
formed by the subglacial geothermal activity beneath
K6 includes the 1918 tephra observed at greatest
depth (∼460 m) southeast of the cauldron. This de-
pression becomes shallower, reaching a saddle point
∼1.4 km southeast of K6 centre (marked C in Figure
11). To understand the formation of the saddle point,
results from recent modelling work, carried out to es-
timate the power of the geothermal areas beneath the
ice cauldrons of Mýrdalsjökull in 2016–2019 (Jarosch
et al., 2020) were inspected. In the modelling, 3D-
velocity fields for the slow, gravity driven flow of
ice were computed with the stationary incompressible
Stokes equations (see e.g. Jarosch, 2008 for details).
The modelling used the bedrock DEM, presented here
(without improvements from the 2021 RES-survey
around the 1918 eruption site), and annual surface
DEMs in 2016–2018 (from Pléiades autumn images)
as inputs. The rate factor A in Glen’s flow law (Glen,
1955) was tuned by fitting the model output veloc-
ity with results from GNSS stations at various loca-
tion including stations at K6 centre in 2018, resulting
in A=2.6× 1024 Pa−3 s−1. A textbook value of n=3
(e.g. Cuffey and Paterson, 2010) was assumed for the
nonlinearity exponent in Glen’s flow law. From the
velocity field in September 2016, modelled using sur-
face DEM from the same month, we obtain horizontal
surface velocity between 6 and 9 m a−1 along an ap-
proximate flow line between K6 centre and the saddle
point C. Assuming that this velocity is representative
for the ice motion at this location since 1918, this indi-
cates that tephra now at C fell on the glacier surface in
1918 only 600–900 m closer to K6 or at least 500 m
southeast from the cauldron centre. This tephra did
therefore not pass the area where most of the down-
JÖKULL No. 71, 2021 61