Jökull - 01.12.1980, Qupperneq 26

to the samples collected in the jökulhlaups
before 1972.
Table 1 lists the available data on the
amount of suspended sediment load in the
Skeiðará jökulhlaups. The calculations are
based on S. Rist’s discharge measurements
(Rist 1955, 1973, 1976) except for the jökul-
hlaup of \938(Thorarinsson 1974). In the jökul-
hlaups prior to 1972, in which the discharge of
Skeiðará was not known, the total sediment
quantity was calculated on the basis of the
estimated total discharge of each Grímsvötn
jökulhlaup.
For the jökulhlaups of 1954 and 1976 the
total suspended sediment was calculated in
two ways. First, by multiplying the total dis-
charge of the jökulhlaup by the average sus-
pended sediment concentration; this is a fairly
inaccurate method. Secondly, by multiplying
the average discharge for each twenty-four
hours by the sediment concentration in the
sample taken on that particular day and
summing the obtained vaiues. For a few days
at the beginning and end of each jökulhlaup
the sediment load had to be estimated. This
does not cause much error because the greatest
part of the suspended load was transported in
only a few days; for example, half the load was
carried in just three days. The second method
was used in 1976 for the Sl- and F-samples as
they were then most numerous. The difference
between the results arrived at by the two
alternative methods is comparatively small,
ranging between 0.5—1.5 million tons. The
values shown in Table 1 are obtained by the
latter method.
In calculating the suspended sediment load
of the jökulhlaups of 1938, 1960, 1965 and
1972 the former method was used exclusively.
In 1938 only two samples were taken. No
hydrographs are available for the jökulhlaups
of 1960 and 1965. In 1972 samples were col-
lected in relatively many places having dif-
ferent concentration, and the sampling
covered a shorter period at each sampling
station than in 1954 and 1976.
Table 1 shows that in the jökulhlaup of
1976, which is the only flood in which dif-
ferent methods of sampling were applied,
these three alternative types of samples give
highly different quantities of sediment load.
This marked difference due to methods must
he kept in mind when comparing the total
load of individual jökulhlaups as given in
Table 1. The F-samples will not be discussed
further since that sampling method has been
abandoned.
It should also be pointed out that the total
volume of water in the jökulhlaups is fairly
uncertain due to difficult measuring con-
ditions. S. Rist (1955, 1973, 1976) considers the
probable error in the water volume to be 15%
in 1976 and 20% in 1954 and 1972. The error
in the other jökulhlaups can be assumed to be
still higher.
TEMPORAL AND SPATIAL
DISTRIBUTION OF SEDIMENT LOAD
In Fig. 3 sediment rating for the Sl-samples
taken from the bridge in 1976 and S3-samples
taken from the riverbank is compared. The
best fit for a curve of the form
y = A xm (1)
was calculated. Here y is sediment load and x
discharge. A and m are constants to be fitted
by linear regression. Both types of samples
show close correlation between discharge and
sediment load. The correleation coefficient for
the rating curve based on Sl-samples is 0.994
but for S3-samples it is 0.990. The Sl-samples
yield considerably greater sediment load and
indicate less spatial distribution than the S3-
samples but this difference mainly applies to
the coarser part of sediment. The rating curves
almost coincide for the finer grain sizes.
The sediment discharge rating on Fig. 4 is
plotted in order to compare the S3-samples
taken from land at the three different samp-
ling localities (at the glacier outlet, at the
bridge and 10 km downstream of the bridge).
The plots show decreasing sediment load
transport with distance downstream on the
sandur plain. This especially applies to the
coarse sediment load at the lowest sampling
24 JÖKULL 30. ÁR