Jökull - 01.01.2005, Blaðsíða 83

A late Holocene jökulhlaup, Markarfljót, Iceland: nature and impacts
abandoned flood channels and back into the gorge
further downstream. As it reached the Markarfljót
sandur, at the end of the gorge system by Þórsmörk,
the flow was focussed in the centre of the valley, not
spreading out to the low slopes along the valley mar-
gins at Langanes or the northern Markarfljót banks.
This flood became channelled as it reached Fljótshlíð,
with tephra-rich flows following the general line of
the Þverá, swamping the trees at Drumbabót. It seems
likely that over Landeyjar the flow was confined to
the main river channels and if any flow extended over
the wider areas it was at a lower velocity, mainly wa-
ter, carrying only finer sediment. Haraldsson (1981;
43) gives a maximum age of 1350±65 14C yr BP (U-
4298, 1390–1130 cal BP) close to river channels in
Landeyjar which may have been deposited by over-
bank flow or wind displacement of fine material from
channel deposits. This date compares well with the
Drumbabót tree date presented here (1230±35 14C yr
BP, AA-48027, 1270–1060 cal BP, Smith 2004) and
suggests that both of these deposits may have been
emplaced during the same flood event.
IMPLICATIONS
Consideration of the Drumbabót story presents impli-
cations for understanding the environmental impact
of medium/small scale flood events and the scale of
floods reflected in the Holocene sedimentary record.
This flood at 1230±35 14C yr BP (Smith 2004)
deposited multiple metres of flood sediment within
the channel of the Þverá covering much of the sandur
area with a thin coating of silt and sand. Woodland
was swamped and killed and there is no evidence of
reestablishment of similar mature woods. Recovery
of vegetation on mudflow and pyroclastic surfaces is
related to substrate stability, presence of sufficient or-
ganic content and nutrients in the sediment and avail-
ability of soil moisture in the period of seedling es-
tablishment (Frenzen et al. 1986). Although the main
tephra-rich sand and gravel deposits would be free-
draining and less suitable for colonisation, silt-rich
low-flow or windblown deposits outside of the main
channels may have been nutrient rich and thus have
rapidly become revegetated and stable.
Despite causing a major influx of sediment into
the drainage network of the lowlands the Drumba-
bót c. 1230 14C yr BP flood did not cause any ma-
jor channel changes. Major changes in the drainage
pattern in the lower part of the region did not occur
until 500 years later, around 1200 AD (Haraldsson
1981). Then, probably in a single event, the exist-
ing pattern with numerous (at least seven) relatively
small courses on the plain was abruptly changedwhen
all courses were combined in a single channel on the
eastern part, and peat deposition started in the aban-
doned courses. The cause of this event is unknown,
but may be related to a climate change and increased
glaciofluvial activity. So far no sedimentary or his-
torical records have been found for a volcanogenic
jökulhlaup at this time.
The majority of the Holocene floods identified in
the Markarfljót valley left traces of sediment along
the valley margins, filled the approximately 5 km
wide valley in the sandur area and filled and over-
topped the gorge upstream (Smith 2004, Larsen et al.
2005). Gröndal et al. (2005) state that such floods
had a discharge, if water-dominated, of over around
100–300,000 m3/sec. Since the Drumbabót flood has
mainly been identified based on evidence found in
the sandur area and did not seem to extend so far
up the valley sides we infer that it was significantly
smaller than other floods recorded (less than 100,000
m3/sec if expressed in the same terms), and has only
been recognised because it is a relatively recent event.
Such medium and small scale events are not repre-
sented in the older Holocene sedimentary records,
since their deposits have either been scoured away or
buried by subsequent fluvial and flood activity. This
lack of evidence however cannot be interpreted as a
lack of jökulhlaup activity in the Markarfljót between
the biggest Holocene events, meaning that frequen-
cies calculated based on sedimentary records must be
understood as minimum estimates only. As can be
seen from the Drumbabót story presented here, even
smaller scale flood events have overtopped the gorge,
washed over the banks of the main river channels on
Landeyjar and been catastrophic enough to destroy an
entire mature birch wood. Indeed, neither Haraldsson
(1981, 40; 1993) nor Påhlsson (1981; 60) found any
signs of organic deposits older than around 2000 14C
JÖKULL No. 55 83