Jökull - 01.01.2021, Qupperneq 59

Bedrock and tephra layer topography within the Katla caldera
lighted by the new bedrock DEM. Topographic fea-
tures such as rows of peaks with north-north-westerly
and northerly directions were visible in the old DEM
(Figure 2b). These features, which may be formed
during fissure eruptions within the caldera are re-
vealed in more detail in the new DEM (Figure 9).
Five ice cauldrons are clustered along one of these
lines. The caldera rim has approximately elliptical
form with its major axis aligned from south-east to
north-west splitting the caldera into the north-eastern
and the south-western halves. The more rugged and
elevated terrain in the south-western half, compared
with the substantially deeper north-eastern half, was
attributed to higher eruption rate, since the volcano
became ice covered (Björnsson et al., 2000). These
topographic characteristics, now even more clearly
defined by the new bedrock DEM, also raise the ques-
tion whether the Katla caldera is a single caldera for-
mation or if smaller caldera formations exist within
the main caldera. The most conspicuous candidate is
a large depression defined by the main north caldera
rim and the steep mountainside with up to 200 m
topographic relief that is aligned east-south-east and
crosses the centre of the main caldera. The sug-
gested sub-caldera, outlined in Figure 9, has an area
of ∼45 km2, corresponding to almost half the main
caldera area. If we assume this sub-caldera was
formed during a caldera collapse event (or repeated
events) overprinting the large main caldera the vol-
ume of the sub-caldera formation can be estimated
by studying how much the mean elevation of the sub-
caldera floor (∼820 m asl) differ from the mean eleva-
tion of the main caldera floor when the suggested sub-
caldera is excluded (∼930 m asl). This correspond to
a volume of 5 km3. The suggested sub-caldera for-
mation is situated above the Katla magma chamber
as inferred from seismic undershooting (Guðmunds-
son et al., 1994). Another possible sub-caldera forma-
tion is ∼5 km2 depression between K6, K7 and K19
slightly south of the caldera centre (Figure 9). It is
∼250 m deep relative to its surroundings and is close
to a suggested location of the 1755 Katla eruption site
(Björnsson et al., 2000). Thirteen out of the 20 estab-
lished ice cauldron locations are close to the rims of
these two suggested caldera formations.
The new bedrock DEM also reveals some pre-
viously unknown topographic features. The most
prominent of those are sharp depressions near
Goðabunga at the west-rim of the caldera (area out-
lined with dashed white line in Figure 9). The de-
pressions are up to 250 m deep and partly surrounded
with 100–200 m high cliff faces. These structures,
which do not align with likely glacier motion along
the glacier surface slope, do not resemble features
carved by glacier erosion. It is more likely that they
are either a complex set of ridges formed by eruptions
beneath the glacier or collapse structures. To better
understand the nature of these features a denser RES-
survey of this area is required for further improving
the bedrock DEM of this area, ideally with 3D migra-
tion since these steep wall structures are hard to map
accurately from 2D migrated RES-data. An example
of topographic mapping of similar structures is shown
in Figure 10 where 3D migration reveals a ∼200 m
high cliff face of a mountain east of K9. This cliff is
not formed by glacier erosion since it opposes the ice
flow from a higher part of the glacier south-west of the
mountain. This mountain, which is the most promi-
nent feature in a previously mentioned row of peaks
(Figure 9) is likely formed by an eruption and further
carved by ice-volcano interactions. This area was ini-
tially surveyed with dense profiling allowing 3D mi-
gration as it was assumed to be likely location of the
1918 eruption. However, recent work (Gudmundsson
et al., 2021; Larsen and Högnadóttir, 2021) locates
the main eruption site of 1918 to be ∼1 km east of
this area, which was densely surveyed with RES in
May 2021 (see Addendum).
DEMs from 3D versus 2D migrated RES-data
The DEMs obtained from 3D migrated RES-data
highlights the limitation of 2D migrated data in areas
of steep and rugged bedrock topography. Such a com-
parison is given in Figure 5 showing the difference be-
tween results from 2D migrated RES-survey carried
out in the spring 2016 around K6 and from 3D mi-
grated survey carried out in spring and autumn 2017.
The 2D migrated bed traces are on average ∼10 m
higher, with ∼20 m standard deviation of the eleva-
tion difference, when compared with the 3D migrated
DEM (Figure 5c). Similarly, when a bedrock DEM
JÖKULL No. 71, 2021 57