×

You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Title
Proceedings of the Symposium on Global and Environmental Monitoring

The change of spectral signatures of
beech and spruce by forest damage
W. Kirchhof & H. Hoffmann
DLR - German Aerospace Research Establishment,
Institute for Optoelectronics, D-8031 Oberpfaffenhofen,
F.R. Germany
Abstract
In the Federal Republic of Germany
multispectral scanner techniques were
developed for forest damage
classification and mapping. To
improve our understanding of spectral
signature changes by forest damage
and to optimize damage identification
and separation high resolution
spectroradiometer measurements of
tree components were performed in the
laboratory, simulating scanner data
acquisition. The measurements of
healthy and damaged beech and spruce
branches proved the influence of
primary effects of damage on spectral
reflectance in the 0.49 - 2.50 pm
region. In the visible discolouration
influences largely the course of
spectral signatures, causes red shift
of the reflexion peak and a variation
of shape at the beginning to the NIR
plateau. In the IR reflectance
decreases with increasing damage
level for beech. For spruce branches
the NIR plateau level reduces also
with increasing damage. For
wavelengths greater 1,4 pm the curves
raise with higher damage level.
Keywords: Forest damage, spectral
signature, beech, spruce,
remote sensing, classifi
cation
0 Introduction
During the last decade in the Federal
Republik of Germany and adjacent
central European countries severe
forest damage has been observed. This
damage comprises large regions and
varies largely. It seems neither
caused by natural hazards like storms
or climatic anomalies or of "classic"
biotic or abiotic origin.
Damage is optically visible by the
loss of leaves or needles, in many
cases by their discolouration (yello
wing) , the thinning of crowns and
anomalous branching.
Damage symptoms become at first
visible in the crown section. Damage
measuring systems, therefore should
preferably gather data with the view
at the top of trees. Aerial photo
graphs or multispectral scanner data
from airborne or satellite plattforms
fulfill these requirements.
At the beginning of forest damage
mainly fir and spruce trees were
affected, Hermann et al., 1988, but
since 1983 steep damage increase for
deciduous trees was recorded, Land
auer et al., 1989. Following the
results of the 1987 forest damage
statistics for the Federal Requblic
of Germany 65,7 percent of beech and
64.5 percent of oak trees, Anon.,
1987, were already damaged. 39
percent of forests are deciduous, 61
percent coniferous forests, total
damage amounts to 50,2 percent in
1984 52,3 in 1987 and 52,4 in 1988.
Forest damage demanded inventories of
its distribution and change over
large areas in support of further
studies, to find the reasons for the
disease. Therefore, in 1987- 1989 the
German Ministry for Research and
Technology (BMFT), Bonn, supported a
cooperative research project for the
development of methods for the
detection, classification and mapping
of forest damage in larger areas of
Germany by the use of remotely sensed
multispectral data.
The main research objective was the
development and to some extent the
verification of new multispectral
scanner techniques including image
processing. This approach required a
new understanding of the information
content of multispectral forest data
in respect to damage classes of the
same species. Five university
institutes, two private firms and the
German Aerospace Research Establish
ment (DLR) took part in the project.
The test areas were selected for
forest damage investigations on fir,
spruce, pine, beech and oak trees and
stands. A description of the project
and results obtained is given in
Landauer et al., 1989.
At the beginning of the project
forest damage inventories were based
on interpretation of CIR - aerial
photographs. By the use of standard
evalution methods and a new forest
damage classification key comparable
results could be reached for sample
areas in different regions and at
variable data acquisition times. The
main information about damage or
disease was derived from colour, but
this source was varying rather widely
i.e. with film type, development
process and flight mission. It was
one of the major results of a
specialist group for photo
interpretation of forest damage,
summarized by Kenneweg, 1989, that' a
general interpretation key,
applicable for different regions and
flight times is possible and
structural elements of tree crowns
and their reproduction in aerial
photographs rather than colours are
used as the main source of
information for identifying different
damage classes". in ’ addition
discolouration by forest damage
can