Skógræktarritið - 15.05.2001, Page 153

SKÓGRÆKT HANDAN SKÓGARMARKA / NSSE
ULRIKA DAHLBERG AND COOMAREN P. VENCATASAWMY
Biomass of mountain vegetation
in optical satellite data
SAMANTEKT
Lífmassi gegnir mikilvægu hlutverki í svæðisbundinni og hnattrænni
kolefnishringrás. Þörf er á að meta lífmassa gróðurs í fjalllendi Skandi-
navíu. Þess vegna beindist þessi forkönnun að undirbúningi hugsan-
Iegrar notkunar gervihnattargagna til að meta lífmassa fjallagróðurs.
Stuðst var við eina IRS LISS III gervihnattarmynd frá I. september
1998 og gögn af athugunarstað á fjallasvæði fyrir norðan Svíþjóð
|68°20' N, I8°50' E). Aðhvarfsfall var metið með IRS LISS III gögnum og
gögnum af jörð úr reitum innan athugunarsvæðisins. Landslag og Iág
sólarhæð í fjöllum á háum breiddargráðum geta orsakað mun á birtu í
halla sem snýr í mismunandi áttir, og voru þessir þættir útskýrðir í
aðhvarfsgreiningunni. Fjögur bönd og landslagsbreyturnar sin(halli),
sin(átt), samverkun milli þeirra (sin(halli) x sin(átt)), og hæð voru
prófuð sem skýribreytur með lífmassa sem svarbreytu. Einungis bönd
3 og 4 voru marktæk, það kom á óvart að landslagsþreytur þættu ekki
aðhvarfsgreininguna. Þetta ersennilega vegna þess að ekki fengust
nægilega margar hallagerðir fyrir athugunarsvæðið til að marktækar
niðurstöður fengjust. )afnframt getur landslag haft þein áhrif á magn
lífmassa, sem væri nægilega stórtil að gera samverkunina ómarktæka.
Vegetation biomass plays an
important role in regional and
global carbon cycles. For exam-
ple, climate change affects the
vegetation cover of the earth,
which influences the amount of
CO^ stored in plants and soil.
These effects are especially large
in sensitive areas such as the cir-
cumpolar ecosystem. There is
therefore a need for improved
estimates of biomass of forests
at a global scale in such areas
(Brown etal. 1999). However,
there are not many studies that
estimate biomass of the moun-
tain ecosystems in Scandinavia.
Due to access difficulties and
high variability of the ecosystem
in those areas, field data mea-
surements are not practical. A
viable alternative for such esti-
mations is the use of optical
satellite remote sensing data.
Several studies have already pro-
posed using digital satellite data
for the assessment of forest para-
meters such as biomass (Hagner
1990, Anderson et al. 1993, Tiwari
1994, Fazakas et al. 1999, Reese
and Nilsson 1999, Steininger
2000). Remote sensing is sug-
gested as the best method to
estimate forest parameters at a
global or regional level (Koch
1996).
Although there exists many
approaches to estimate biomass
from satellite images these
methods need to be evaluated
for mountainous vegetation,
especially because of the slopes
and elevation characteristics of
the Scandinavian Mountains and
the relatively low sun angles in
the area. Regression models are
the most commonly used
method for this purpose, and the
individual wavelength bands
from satellite images, or vegeta-
tion indices derived from a com-
bination of wavelength bands,
are usually used as explanatory
variables (Anderson etal. 1993,
Hagner 1990, Salvadorand Pons
1998, Steininger 2000). The
results, however, have differed.
Steininger (2000) found a signifi-
cant linear relationship between
mid-infrared reflectance, derived
from band 5 of Landsat TM-data,
and biomass in Brazilian stands.
However, Salvador and Ponds
(1998) found poor fits of simple
or multiple regression models of
forest variables with the individ-
ual bands of Landsat TM and the
normalised difference vegetation
index (NDVl) as explanatory vari-
ables. Anderson et al. (1993) did
not find a good relationship
between biomass from sample
points and a number of vegeta-
tion indices such as difference
(DVI), ratio (RVl) or normalised
difference (NDVI), derived from
SKÓGRÆKTARRITIÐ 2001 l.tbl.
151