Jökull - 01.01.2020, Blaðsíða 83
Guðmundsson and Björnsson
Figure 6. Glacier elevation along the lateral margins of the North (blue triangles) and South (yellow circles)
Kvískerjajöklar at LIAmax (a) and in 1930 (b) with respect to the elevation in 2011. The least-squares linear
relationship between 650 m a.s.l. and 1600 m a.s.l., assuming insignificant elevation changes at higher eleva-
tions, is shown with a dotted line. – Yfirborðshæð á jöðrum nyrðri (bláir þríhyrningar) og syðri (gulir hringir)
Kvískerjajökla, í hámarksstöðu á litlu ísöld (t.h.) og 1930 (t.v.) samanborið við árið 2011.
ablation area. During the LIAmax it extended be-
yond a steep vertical drop of the narrow Múlagljúfur
gorge. The South glacier was modelled by four flow
lines, overflowing the Rótarfjallsgljúfur gorge during
the LIA and merging with the North glacier farther
downstream. Figure 7a–b shows the surface lowering
over the period from 1880 to 2011, along the profiles
P1–P2 for the North glacier and P3–P4 for the South
glacier (see Figure 3). Area and volume changes
are given in Table 1. The computed DEMs did not
account for rock buttresses that shape the northern
rim of Múlagljúfur, as described by historical records
(Pálsson, 1945; Henderson, 1957; Thoroddsen, 1959;
Björnsson, 1998). The final DEMs of the upper and
lower parts of the glacier were merged at an elevation
of ∼600 m with some smoothing of the ice-surface
geometry at the coalescence zone.
Uncertainty of the DEMs
The DEMs of the LIAmax, 1880s and 1930 are partly
constructed by subjective assessments and hence the
uncertainty in elevation is difficult to quantify. In the
elevation range 650–1600 m a.s.l. (∼6 km2) we as-
sume an uncertainty of ±10 in the elevation. The
altitude of the LIA ice surface near the lateral ice
margins, estimated from geomorphological evidence,
is assumed accurate to ±5 m. The elevation error
in the 2011 lidar DEM is <0.5 m. This is much
less than other sources of error and has negligible ef-
fect on the final error estimate. In the range 260–
650 m a.s.l. (∼3 km2) we estimate the error to be
±20 m. In the lower part of the 1930 glacier area,
outside the 2011 DEM (∼3 km2) where the eleva-
tion is based on the GlaRe model, we estimate an
error of ±30 m, mainly due to misaligned contours
and near-margin connection to contours outside the
glacier. Joining the two DEMs will also introduces
some error. Volume changes were calculated by sub-
tracting DEMs in Surfer. The uncertainties of the vol-
ume change between LIAmax and 1880 DEM is esti-
mated as ±0.15 km3 and for the period 2011 and 1930
DEMs as ±0.12 km3 (error contributions from differ-
80 JÖKULL No. 70, 2020