Jökull - 31.12.2001, Side 16

Jukka Kayhkö et al.
type and amount of vegetation cover. Due to the low
spatial resolution of the TM sensor (30x30m), and
the clustering method used, it was not always pos-
sible to satisfactorily distinguish between land cover
types. Problematic land cover types - i.e. those not
clustered in a consistent way - included various mix-
tures of sediment and lavas. Small errors can therefore
be detected on the classified image.
The Landsat TM mapping method used here, and
our raster-format map product, offer the opportunity
to monitor geomorphological processes in this area by
regularly repeating the classification process with re-
cently acquired data. The use of multitemporal data,
however, would call for additional data manipulation,
e.g. atmospheric correction, prior to detailed compar-
isons.
CONCLUSIONS
1. Six lava cover types and nine sediment cover
types were described and mapped by a Land-
sat TM-based classification of the study area.
The most abundant lava classes were rough
lava mixed with sediments and rough lava
with scattered vegetation, with ca. 10% areal
coverage each. The most abundant sediment
classes were medium and coarse sediments and
alluvial deposits, which dominate the sandur in
front of Dyngjujökull.
2. Streamlined sedimentary deposits in the west-
ern part of the study area (the three Frambruni
sand lobes and the Fremrinámar sand stretch
east of Bláfjall and Búrfell) suggest an aeolian
sand transport process towards the north/north-
east.
3. Major jökulhlaup deposits with adjacent dune
fields were detected along the Jökulsá á Fjöll-
um at Möðrudalur and Grímsstaðir, suggesting
that jökulhlaups serve as a potential triggering
mechanism for soil erosion.
4. The Landsat TM-based mapping method de-
scribed here could be used to monitor this re-
gion’s geomorphological processes by repeat-
ing the classification process with recently ac-
quired data.
5. Due to the spatial resolution of the TM and
the clustering method used here, it was not al-
ways possible to distinguish between land cover
types - this was particularly true for types that
form a continuum in the field.
Acknowledgements
The Postgraduate Research Institute for Sedimento-
logy (PRIS) and the Department of Geography at the
University of Reading, UK provided the infrastruc-
ture for the early phase of this study. The Depart-
ment of Geography, University of Helsinki, and the
Laboratory of Computer Cartography, University of
Turku, kindly supplied the equipment and facilities re-
quired to finish the study. Prof. Peter Worsley, Prof.
Ken Pye from PRIS, and Prof. Matti Seppala from
the Department of Geography, University of Helsinki,
are thanked for supervising JK’s PhD studies. Dr.
Geoffrey Griffiths from the Department of Geogra-
phy, University of Reading, offered kind advice in re-
mote sensing applications. We want to thank Dr. Rob
Bryant, Ms. Saara Kinnunen and Mr. Sakari Palo for
invaluable assistance in the field. We are grateful to
Dr. Árni Einarsson, Dr. Jón Ólafsson, and all other re-
searchers at the Mývatn Biological Research Station
for hospitality and good spirit. Dr. Ólafur Amalds
from RALA has offered invaluable help and fruitful
comments in many occasions. We are grateful for
the financial support offered by the following insti-
tutions, without whom it would not have been pos-
sible to complete the study: the Academy of Finland
(projects 1012238 and 2247), Emil Aaltonen Foun-
dation, the British Council, the CIMO, the Finnish
Cultural Foundation, the Iceland- Finland Cultural
Foundation, Alfred Kordelin Foundation, Maj and
Tor Nessling Foundation, the Nordic Arctic Research
Programme (NARP), the Overseas Research Students
Awards Scheme (ORSAS) UK, the Sohlberg Delega-
tion of the Finnish Science Society, the University of
Helsinki, and Jenny and Antti Wihuri Foundation. We
thank the Science Council of Iceland for granting the
research permit. PRIS contribution no. 836.
14 JÖKULLNo. 51