Jökull - 01.01.2014, Qupperneq 51

Magnetostratigraphy, K-AR dating and erosion history of Hafrafell, SE–Iceland
at the 2 standard deviation (95% confidence) level.
These age determinations were originally reported in
an appendix to Helgason and Duncan (2001). Here
we present revised calculated ages in Table 2, using
the decay and abundance constants recommended by
Min et al. (2000) to conform with geologic time scale
divisions of Gradstein et al. (2012).
DISCUSSION
Stratigraphic correlation with the geomagnetic
time scale
We next correlate our results from K-Ar dating and
paleomagnetic polarity determinations on Hafrafell
units with the geomagnetic time scale (Gradstein et
al., 2012) as shown on Figure 8. The suggested cor-
relation is based on data from profiles HL, HP1, HP3,
U and T (Figures 4 and 9 and Table 2). The K-Ar age
of each dated unit is shown at its stratigraphic posi-
tion together with the stratigraphic position of erosion
surfaces. The very lowest lavas sampled for paleo-
magnetic signature in section HP at the south end of
Hafrafell (units HP1-1 and HLX) differ from the lavas
next above in that one is normal while the other is R-
transitional (shown as Nt on Figure 8). This suggests
that the lavas at the very base of Hafrafell may belong
to the Cochiti normal subchron (C3n.1n; 4.187–4.300
Myr). The lowest unit in section HL (Figure 4), that
is normally magnetized, was dated at 3.92±0.06 Ma
and correlates with the Cochiti normal sub-chron.
Next on top of unit HL1 begins a sequence with
three main magnetic polarity intervals or upwards
from R1-N1-R2 (Figure 8). The 829-m-thick R2-
sequence has been dated near its base at 2.35±0.22
Ma (unit HZ), which identifies the R2-lava sequence
as lower Matuyama (C2r; 1.945–2.581 Myr). Unit
HZ, from an 85-m-thick subglacially erupted ridge,
formation (HF10), generated relief. The first lava to
bank up against it during the next interglacial stage
showed clear lava ponding as is shown on Figure 2
with a blue line below a thick basalt lava flow beside
formation HF10. The 234-m-thick N1-lava sequence
below, for which we report an age of 3.20±0.09 Ma,
is correlated with the Gauss interval (C2An; 2.581–
3.596 Myr). The correlation suggests that the only
part of Gauss time represented by Hafrafell strata
is the N1 sequence that we correlate with C2An.3n
(3.330–3.596 Ma). Unit HL27, in the upper part of
N1 has an age of 3.20±0.09 Ma. Above dated unit
HL27 in section HK are some 9 N-lavas (Figure 9).
Thus lack of strata during Gauss in Hafrafell supports
a hiatus during the period 2.581 to 3.330 Myr as sug-
gested on Figure 8.
Higher up the section, in the Hafrafell valley fill-
ing, a number of reversals and brief magnetic events
are recorded. Here the transitions are R2-N2-R3-N3-
R4 (Figure 8). This sequence is older than Brunhes
but presumably not far below the Brunhes/Matuyama
boundary. A unit from R4 was dated at 1.69±0.29
Ma suggesting that the short N2 and N3 intervals cor-
relate with the Olduvai subchron (1.778–1.945 Myr).
Toward the top of the section all units are normally
magnetized (N4), e.g. groups H7 and H8, and clearly
of Brunhes age or younger than 781 kyr. The re-
ported age of 215±12 ka for a lava flow from for-
mation HF39 (group H8) allows this formation to be
correlated with the third last interglacial (Mindel-Riss
for the Alps; Holsteinian for N-Europe) and MIS 6
at about 191 ka (Gradstein et al., 2012; Lisiecki and
Raymo, 2005).
Stratigraphic division into groups H1 to H8
In order to trace the erosion history and landscape
development at Hafrafell we have combined the 39
mapped rock formations into 8 groups (H1 to H8),
from oldest to youngest. The distribution of the
groups is presented in Figure 5 along with the 12 ero-
sion surfaces, HR1–HR12. Of particular interest is
the thick lava group, H5, into which a valley was in-
cised, that probably represents an established valley
network. We refer to this, at least 260-m-deep, de-
pression as the Hafrafell valley. The valley sides con-
sist of at least 739-m-thick lava sequence that formed
during a relatively short interval, 1.945–2.581 Ma, or
0.64 Myr. The timing of the valley formation can be
narrowed down to being older than the Olduvai sub-
chron, as the stratigraphically lowest valley-infilling
lavas are reversely magnetized and predate the Oldu-
vai subchron. An age of 2 Ma would therefore be rea-
sonable for the valley.
JÖKULL No. 64, 2014 51