Jökull - 01.01.2004, Blaðsíða 28

Eyjólfur Magnússon et al.
with full weight (Pi < ρigH) on the discharge tunnel,
the counteracting strength of the tunnel itself has thus
become important.
APPLICATION TO ICE FLOW AND
MASS BALANCE STUDIES
Location of ice divides
The gravitational force drives glacier flow. On a re-
gional scale, flow lines and ice divides can be derived
from the glacier surface DEM. The surface EMISAR
DEM used in this study was sub-sampled down to
100x100 m. It was filtered prior to the plotting of the
ice divides with a same kind of filter as used when
calculating the water potential. However, a filter with
the width from −σ to σ equal the ice thickness leaves
out far too much surface details for drawing the ice
divides. Considering movements of large ice masses
only large-scale surface changes matter ( e.g. Patter-
son, 1994). Therefore filters with width from −σ to
σ equal to n times the ice thickness, where n = 2,3,4
or 5, were applied and tested. The same procedure
was used to for drawing the ice divides as for the wa-
ter divides. Mountains or other clear landforms could
normally be used for defining the point at the margin
from which the divides were traced up the glacier.
For n=2 the ice divides for Síðujökull, Skaftár-
jökull and Tungnaárjökull are uneven and the narrow
ice basin of Skaftárjökull is reduced to nothing be-
fore reaching the accumulation area. Applying n=3,
n=4 and n=5 gives quite similar results except the fil-
ter with n=5 almost smoothed out the Skaftár caul-
drons leaving them with much smaller ice basin than
expected. The ice divides using n=3 and n=4 are al-
most identical. Here the divides with n=3 are shown
(Figure 8). The corresponding area values are given
in Table 2 in the Appendix.
As an alternative to the Gaussian filters used in
this study triangular or exponential filters may be used
for filtering surface DEMs prior to estimating glacier
flow lines (Kamb and Echelmayer, 1986). Since vari-
ous sizes of Gaussian filters gave such comparable re-
sults, the alternative filters would probably give simi-
lar results even though the size of the filters used for
the presented ice-divide might be different.
Ice divides in 1998, compared with results based
on DEMs in the 1980s (Björnsson et al., 1988b,
1992b; Defence Mapping Agency and the Icelandic
Geodetic Survey, edition 1-DMA, series 761) show
large deviations around the 1996 eruption site, Gjálp
(Figure 8). The depression around Gjálp generates
a significant decrease in the ice basin of the Eastern
Skaftá cauldron. There is also a considerable change
in the locations of the ice divides between Síðu-
jökull and Skeiðarárjökull caused by the last surge of
Síðujökull in 1993–1994. This surge probably also
changed the location of ice divides between the up-
permost parts of Síðujökull and Skaftárjökull.
The area distribution of W-Vatnajökull outlets
Having defined the present ice divides on W-
Vatnajökull we can define the area distribution of its
major outlets. Figure 9 shows the accumulation area
ratio (AAR; the ratio of accumulation area to the to-
tal area of the outlet) as function of equilibrium line
altitude (ELA) for 5 outlets of W-Vatnajökull, accord-
ing to the present ice divides and the elevation distri-
bution for the EMISAR DEM in 1998 and the DMA
maps in the mid 1980s (Defence Mapping Agency and
the Icelandic Geodetic Survey, edition 1-DMA, series
761). Based on experience from numerous field sur-
veys and comparison with the EMISAR DEM, which
shows very consistent elevation difference, the quali-
ties of the DMA maps for these five outlets are con-
sidered acceptable except the highest part of Köldu-
kvíslarjökull.
Sylgjujökull is most sensitive to changes in the
equilibrium line elevation since the main part of that
outlet is situated in a narrower elevation range than
the other outlets. The AAR ratios are similar for the
other outlets. The actual ELA on Tungnaárjökull and
Köldukvíslarjökull, on which the balance is measured
directly, is posted onto the graphs from 1998 (Figure
9) for most of the years 1992–2001 (Björnsson et al.,
1998, 2002). The earliest AAR values, for Köldu-
kvíslarjökull and Tungnaárjökull should be taken with
caution. The Tungnaárjökull surge in 1994–1995
(Björnsson et al., 2003) altered its elevation distri-
bution significantly causing an underestimate in the
AAR values for 1992–1994. Older estimates for the
AAR on Tungnaárjökull give the values 0.65 in 1992,
28 JÖKULL No. 54