Jökull - 01.01.2021, Síða 62

Magnússon et al.
is much less when compared with unpublished water
divides made by IES-glaciology group, using the old
bedrock DEM (Figure 2b) and the lidar surface DEM
from 2010 (Jóhannesson et al., 2013). The DEM from
2010 both resembles more the 2019 glacier surface
and is much more accurate than the surface DEM used
in Björnsson et al. (2000), which was based on mea-
surements in 1991.
The known geothermally-sustained cauldrons are
all within the drainage basins of the rivers Markarfljót,
Jökulsá á Sólheimasandi or the rivers draining from
Kötlujökull, except K12, which drains to river Leirá
draining from northeast Mýrdalsjökull. All cauldron
locations fall within the same drainage basins for both
the previous (Björnsson et al., 2000 and the unpub-
lished one using the 2010 lidar survey) and the new
delineation, with the exception of K19. This cauldron
now falls within the drainage basin of the river Jök-
ulsá á Sólheimasandi but had previously fallen within
the Kötlujökull drainage basin in previous estimates
of the water divides. It should be noted, however, that
the water divides at this location are very sensitive to
errors in input data and the assumptions made when
estimating the water potential. Hence it is quite uncer-
tain to which drainage basin K19 belongs. The same
applies to K5 and K6 as the estimated drainage route
from these cauldrons passes below K19. However, it
is fairly certain to which drainage basins all the other
cauldrons belong.
The derived water potential has many local min-
ima, which can be viewed as puddles in the poten-
tial, facilitating water accumulation at the bed. These
puddles are typically found beneath the geothermally
sustained cauldrons (Figure 7b). It should however
be noted that potential puddles are also found outside
known cauldron locations. Beneath some of the caul-
drons, known to release water in jökulhlaups, either
no or only small and shallow potential puddles are de-
rived. This even applies for cauldrons where the bed
has been obtained from 3D migrated RES-data, in-
cluding cauldrons K16, K9 and K11 (the applied sur-
face filtering mentioned above only slightly reduces
the size and the depth of the potential puddles). K9
and K11 are located above steep beds, which almost
evens out the effects of adverse surface slope out of
the cauldron despite surface slope, having 10-fold
stronger influence on the potential gradient (∇ϕ) than
the bed slope. In the case of K16 it did not form a
closed surface depression in September 2019, causing
the absence of a potential puddle. Comparison be-
tween DGNSS data obtained during the RES-survey
in May 2019 and the September surface DEM does
however show that water was released from beneath
K16 during the summer, causing up to 10 m lowering
in the cauldron centre, despite the lack of a potential
puddle. This highlights the limitation of using Eq. 1
to locate potential sources of jökulhlaups. These lim-
itations should be studied further by comparing caul-
dron activity and how the estimated water potential
varies with time, both seasonally and over longer time
scales, using various available surface DEMs obtained
since 2010, as well as the new bedrock DEM.
Tephra layers
The maps of ice thickness in 2016 above two tephra
layers (Figure 8) observed at depth within the caldera
should also be considered as one of the main results
of this study. The older tephra layer is only observed
in the northern part of the caldera, within the thick-
est part of the glacier and was found at 390–600 m
depth in 2016. The younger layer at shallower depth is
tephra that fell on the glacier surface during the 1918
eruption (Brandt et al., 2006) and is still stored in the
ice within most of the caldera. The largest exception
is the southeast part of the caldera (Figure 8a). The
gap in the presented map is partly related to lack of
RES-data (the area south of K10 above the trough to-
wards Kötlujökull), but for the main part of the area
surrounding K8, K9, K16 and K17 the tephra layer is
not observed in the RES-data. At the boundary of the
area where the 1918 tephra layer is observed, north of
these cauldrons, the layer dips towards the glacier bed
and can be traced until it is 30–100 m above the bed.
It is therefore likely that for a large proportion of the
area around these cauldrons, where the tephra layer is
not seen in the RES-data, the 1918 tephra layer has
already reached the glacier bed, due to geothermal ac-
tivity and relatively shallow ice. It likely also plays
a role that the 1918 eruption took place in this area
(Gudmundsson et al., 2021) resulting in even thinner
ice right after the eruption. Geothermal activity has
60 JÖKULL No. 71, 2021