Jökull - 01.12.1994, Qupperneq 35

able time span. The high expected uplift rate in SE
Iceland could be verified in 5-10 years. For this pur-
pose a network of 10 GPS-geodetic points was in-
stalled in the area in 1991. The network occupies an
area of about 1500 km2 between the glacier edge and
the coast. It includes a point in Djúpivogur at a dis-
tance of more than 50 km from the glacier, corre-
sponding to a 120 km distance from the glacier center,
i.e. outside the expected area of uplift.
Three Ashtech MD-XII GPS receivers provided
by the British Natural Environment Research Council
were used in the project. These are dual frequency,
C/A code instruments that record carrier wave phase
data from all GPS satellites in sight. The survey took
place during the period July 6 to 19, 1991. Two ses-
sions were usually measured each day, session 0 from
about midnight to about 10 h, and session 1 from
about 15 to about 20 h. One antenna malfunctioned at
Fig. 4. Uplift curves for two values of lithosphere thickness,
10 km (thin line) and 20 km (thick, shaded line), and viscos-
ity of 1019 Pa s. For each case the three curves represent the
calculated uplift 50 years after unloading, 100 years after un-
loading, and the final equilibrium uplift. Here instantaneous
unloading of the load shown in Fig. 2 is assumed (after Sig-
mundsson, 1990).
4. mynd. Landrís reiknað semfall affjarlœgðfrá jökulmiðju
fyrir tvö iíkön með mismunandi plötuþykkt, 10 km (mjó lína)
og 20 km (feit lína). Reiknað er með seigju undirlagsins 10'9
Pa s. Fyrir hvort líkan eru reiknaðar þrjár rislínur, þ.e.
landris eftir 50 ár, 100 ár og endanlegt ris eftir langan tíma.
Gert er ráðfyrir að alltfargið sé tekið afí einu. Takið eftir
að bœði líkönin gefa heildarrís um 1 m fyrirfyrstu 100 árin í
Hornafirði.
the beginning of the survey and was replaced with an-
other of a slightly different design, but from the same
manufacturer. We were assured that no errors would
result from this change. All points were occupied at
least three times. An overview of the sessions and the
point occupation is given in Table 1. The point HOFN
was used as a reference station for most of the ses-
sions. The location of the points is shown in Fig. 1.
Their descriptions are given by Einarsson (1993).
Data were recorded at 5 s intervals during the day ses-
sion and 10 s intervals during the night session. Data
from satellites below 10° elevation angle were not
recorded. No meteorological data were collected.
The data were analysed using version 3.2 of the
Bernese GPS analysis software (Rothacher et al.,
1990) at the University of Durham. Parameters are es-
timated using double differenced phase observations
and five linear combinations of the two carrier phas-
es: the original L1 and L2 carriers, the ionosphere-
free linear combination (L3), the geometry-free com-
bination (L4), and the wide-lane combination (L5).
Station coordinates are solved for, with the possibility
of cycle ambiguity resolution and orbit improvement.
The data from each session were divided into two
and each half session was processed independently.
This increases the number of coordinate results while
the accuracy of each result decreases. This is consid-
ered to give a better estimate of the true errors, since it
provides information on the repeatability of the re-
sults, i.e. RMS scatter about the mean value. The for-
mal errors of the solutions alone greatly underesti-
mate the true errors (Sigmundsson, 1992).
Noise in some of the data caused problems. Data
from session 198.0 were unusable and parts of ses-
sions 194.1 and 195.1 were also discarded. Judged
from the records of the Leirvogur magnetic observa-
tory (Þorsteinn Sæmundsson, pers. comm.), iono-
sphere disturbances are most likely responsible for
these difficulties.
Cycle slip and outlier detection was performed
during parameter estimation. The following steps
were taken in the analysis: An L3 ambiguity-free so-
lution was found, to obtain initial, estimates of the
point coordinates, whith few cm accúracy. Outliers
were removed. All L5 and over 90% of the L1 and L2
JOKULL, No. 44
33