Náttúrufræðingurinn - 2007, Qupperneq 15

Tímarit Hins íslenska náttúrufræðifélags
9. nnjnd. Hélumosavist greindist nokkuð vel viðfjarkönnun enda þótt gróðurþekja sé lítil
og landið líkt melavistum að yfirbragði. - Moss snowbed habitat type could be detected
by remote sensing. Ljósm./photo: Sigmar Metúsalemsson (ágúst 2004).
SUMMARY
Over the last few years the Icelandic
Institute of Natural history has been
developing methods for classification of
Icelandic habitat types. 7’he sampling of
vegetation, birds, land invertebrates and
environmental factors has been carried
out in seven highland areas, resulting in
identification of 26 different habitat
types. The habitat classification has so far
been based on vegetation mapping of the
areas on aerial photographs.
The main objective of the current
study was to investigate if remote sensing
methods could be used in identification
and mapping of different habitat types to
make the work more efficient. A test area
of 350 km2 in northeast of Iceland was
selected for the research and a SPOT5
satellite image from 2002 of the area was
used for the study. Accuracy assessment
was made for habitat classification based
on vegetation mapping and analysis of
the SPOT5 image. For ground-truthing
sampling of 130 tests transects in 11 habi-
tat types was carried out in the study
area. The total area of each habitat type,
classified by each method, was calculated
and compared. Based on the field data, 12
habitat types were classified, but they had
to be condensed to eight classes in order
to be able to compare the classification of
the two different methods. All four bands
of the SPOT5 image were used for both
supervised and unsupervised classifica-
tion. Corrections of the mapping were
made by using both slope and hydrolog-
ical data.
For all the habitat types unsupervised
classification gave higher correlation with
the test data than supervised classifica-
tion. A map of habitat types was
constructed for the area based on tlie
unsupervised classiíication. Overall accu-
racy of the map for the eight classes was
70%, but tlre accuracy of each class varied
from 25-90%. The accuracy of the SPOT5
map was slightly higher than that of a
map based on traditional vegetation
mapping. However, complete identifica-
tion and separation of all the different
habitat types could not be accomplished
by remote sensing. Also habitat types that
are rare or cover small areas may not be
recorded at all by this method. We conclu-
de that remote sensing is a powerful and
useful tool that can be used to classify
major habitat types but it has to be used in
conjunction with other methods.
ÞAKKIR
Rannsókn þessi er hluti af verkefninu „Nýting fjarkönnunar við vistgerða-
flokkun" sem unnið var í samstarfi Náttúrufræðistofnunar íslands,
Háskólasetursins á Homafirði og Landmælinga íslands. Verkefnið var jafn-
framt meistaranámsverkefni Regínu Hreinsdóttur við jarð- og land-
fræðiskor Háskóla íslands. Það var styrkt af RANNÍS, markáætlun um upp-
lýsingatækni og umhverfismál, 2003-2005. Rannveig Ólafsdóttir við
Háskólasetrið á Homafirði, Kolbeinn Ámason á Landmælingum íslands og
Sigmar Metúsalemsson, sérfræðingur við Landbúnaðarháskóla íslands
tóku þátt í verkefninu. Landsvirkjun lét í té hæðarlínu- og vatnafarsgögn
sem notuð voru í verkefninu. Kunnum við þessum aðilum bestu þakkir.
HEIMILDIR
1. Nagendra, H. 2001. Using remote sensing to assess biodiversity.
Intemational Joumal of Remote Sensing -22. 2377-2400.
2. Lubchenco, ]., Olson, A.M., Brubaker, L.B., Carpenter, S.R., Holland,
M.M., Hubell, S.P., Levin, S.A., MacMahon, J.A., Matson, P.A., Melillo,
J.M., Mooney, H.A., Peterson, C.H., Pulliam, H.R., Real, L.A., Regal,
P.J. & Risser, P.G. 1991. The sustainable biosphere initiative: An eco-
logical research agenda. Ecology 72. 371^112.
3. Roughgarden, J., Running, S.W. & Matson, P.A. 1991. What does
remote sensing do for ecology? Ecology 72. 1918-1922.
4. Stoms, D.M. & Estes, L.E. 1993. A remote sensing research agenda for
mapping and monitoring biodiversity. Intemational Joumal of Remote
Sensing 14.1839-1860.
83