Z»nl»
Zenith
à
where:
i and r = denotes
the incident and the
reflected radiation,
respectively
d 0 = the element of
the radiant power
propagating within
the element of solid
angle d
ft = zenith angle
7 = azimuth angle
Fig. 1 (from Boehnel et al 1978) shows the geometry
of reflection:
R(A)
and:
L Ar
L A‘ wL* Cf
Table 3. Damage classes.
Healthy = 0-10% needleloss
Sickly = 11 - 25%
Damaged = 26 - 60% "
Severely damaged = more than 60% "
3.1 The spectral reflectance signature in the visible
region
The reflection curves of healthy spruce, fir trees
and stands differ clearly in the visible region iron
those slightly or severely damaged and from dead Fir,
spruce trees.
Fig. 2-10 show that damaged trees or stands have a
continually higher reflection between the visible
channels 3-7 than the vigorous trees. This is due
to lower chlorophyl amounts which turn the color of
the needles to yellow or brown and therefore a higher
reflection in the visible region.
The data in Fig. 7 represent an average reflection
value of many (3-4) stands.
L Ar = (GV ~ ~ bbl)
L À r • wL = ^ ~ BBL ^ * C// ^ ~ BBL ^ * * ^ 3.2 Spectral reflectance in near infrared
where:
R(^) = spectral reflectance factor
L^ r = object radiance
GV = grey value (digital number)
BEL = blackbody low of the MSS
C = calibration value of the MSS
RL = reference lamp of the MSS
A A = spectral band width
L Ar wl = re ^ erence panel radiance = global radia
tion
Cf = correction factor of reference panel
The Landsat 5 (TM) data is also converted from grey
value (digital numbers) to reflected radiance in
W/m 2 .sr.mic. as follows:
L^ r = (GW - OF) / (GN 4* AA)
where:
L^ r = reflected radiance
GW = grey value (digital number)
CF = detector offset (digital number)
QSI = gain
£ X = spectral band width
3 RESULTS
The test site is mountainous and therefore single
trees, groups of trees or stands will not be illumi
nated equally by the sun's radiation. This means that
the reflected radiance from similar objects in diffe
rent exposition has various intensities. The reflec
ted radiance from a single tree depends on the den
sity of the foliage and in the case of a damaged tree
(loss of needles or leaves) the detected radiance may
be reflected from the soil or ground vegetation
through the tree crown. This is quite often the case
with damaged tree crowns.
Because of the resolution cell size, this study, as
most studies, evaluated the integrated reflected com
ponent of the...ground area which includes tree canopy,
soils, lower vegetation and shadow. All damaged fo
rest stands are located on ridge tops and healthy
stands are generally on the lower slopes and in the
valley. This distribution of stands also affects the
reflected radiance values.
The reflection in this spectral region is indepen
dent of the plant pigmentation. The intensity of the
reflection depends rather on the vitality of the ve
getation, vigorous trees and stands reflect more than
damaged (Fig. 2 - 10), but the reflectance curves
(Fig. 7) from this study show that the damaged stands
have higher reflection in channels 8, 9 and 10.
This spectral reflectance behaviour was not expected.
There are many possible factors which could explain
this phenonena e.g. the data in Fig. 7 are collected
by aircraft at 1000 m altitude above ground, and the
location of the damaged stands are in the middle of
the test strip on the top of the mountain whereas
the healthy stands are downslope and in the valley.
The observation is therefore different, and the re
flected intensity will be different.
Beech trees or stands always reflect more near IR
radiation than the coniferous types (Fig. 2, 5-9).
3.3 Spectral reflectance in middle infrared
Middle IR data are available for altitudes of 1000 m,
3000 m and Landsat 5 (TM) . The reflection in this
region depends on the water content of the vegeta
tion. Damaged vegetation usually has a lower water
content because of decreased evaporation (defolia
tion and dry branches) and therefore reflects more
radiation than healthy vegetation.
Fig. 7 and 10 show clearly the differences in the
reflection between healthy and damaged stands.
4 CONCLUSION
Knowledge about the spectral reflectance signatures
of forest stands (healthy and damaged) using multi-
spectral data is very important for quantification,
classification and monitoring of forest areas which
show various degrees of damage.
The evaluated data in this study which were collected
from different altitudes show clear differences in
spectral Signatures, in the visible near IR and
middle IR regions of the electromagnetic spectrum
especially in the near IR region. That indicates
the possibility for a computer aided classification
of the .forest damage inventory task.
