W sensor
or forest
d. In this
:hniques.
n density
surveys.
ng DSM
cessfully
specially
the trees.
ial of the
lineation
| used to
teristics.
g on the
bout the
ery with
iceborne
'solution
the Im
detailed
r limited
1 change
gorithms
cable to
gies for
ie recent
ientation
7), edge
9) and
-Camera
y image
V was
manual
potential
letection
using a
research
remote
^ forest
different
meras in
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
combination with different analysis methodologies for the
automatic detection of forest attributes (Heurich et al., 2003).
2. MATERIAL
2.1 Study Area
The research was conducted in the Bavarian Forest National
Park which is located in south-eastern Germany along the
border to the Czech Republic (49? 3' 19" N, 13? 12' 9"E).
Within the park three major forest types exist: There are sub
alpine spruce forests with Norway spruce (Picea abies) and
partly Mountain ash (Sorbus aucuparia) above 1100m. Mixed
mountain. forests with Norway spruce, White fir (Abies alba),
European beech (Fagus sylvatica) and Sycamore maple (Acer
pseudoplatanus) can be found on the slopes between 600 and
1100 m Finally, spruce forests with Norway spruce, Mountain
ash and birches (Betula pendula, Betula pubescens) occur in
valley bottoms with wet depressions often evidencing cold air
ponds.
2.2 Field Data
Three sample plots with a size between 20 by 50m and 20 by
100m were selected in the sub alpine spruce and mountain
mixed forest zones. The field data was collected in May and
September 2003. Several tree parameters like the diameter at
breast height, total tree height and starting point of crown were
determined for each tree being higher than 5m. The height
measurements were carried out with the “Vertex” III system
following the definitions of Kramer and Akca (1995). Each
stem position was precisely measured by tachometry and
DGPS. The absolute accuracy was comprehensively checked
and was estimated to 1-2 cm. The data was subsequently
evaluated and visualised with the help of the program package
SILVA 2.2, which is commonly used for forest simulations
(Pretzsch et al. 2002). The mean annual growth in height was
determined for the stands with 20 to 25 cm. Plot Hochwiesel A
(50) is a typical natural Norway Spruce stand of a subalpine
environment. Its average height above see level is 1240 m.
Feistenhäng À (60) is a pure deciduous tree stand dominated by
European beech. It is located in the mountain mixed forest zone
890m above sea level. The size of the plot is 20 by 50m and the
slope is 15.3%. Plot Lärchenberg B (22) is an old growth stand
typical for the mountain mixed forest zone with
plot Spe- N/h hl00 di00 G V D° %
cies a [m] [em] [m [m?
ha] /ha]
22 Picea 45 41.7 77.4 2] 349. 031 30
Fagus 165 295 37.7 13 189 0.46 44
Acer/Tilia 40 28.7 525 9 130 028 27
250 43 668 1.05
SU Picea 60.6 12 472 10 100
N
Do
Un
ID
ON
Un
60 Fagus 260 34.1 53.1 28 470 038 9)
Acer 20 26.1 38.7 2 32 008 .9
280 30. 502. 0.97
Table 1: Forest characteristics of the sample plots (N/ha:
number of trees per ha, h100: dominant height
(average height of the 100 tallest trees), d100:
dominant diameter, G: basal area, V: growing stock,
D°: stock density).
European beach, Norway spruce and Sycamore maple located
885m above sea level. The size of the plot is 20 by 100 and the
slope 1s 29.8%.
2.3 Aerial Data
2.3.1 Digital Modular Camera (DMC): The DMC flights
were performed in end of June 2003 in four separated blocks
The flying height was 960m resulting in a photo scale of
[:9.000 and a GSD of 11 cm. The images were captured with
70% end lap and 50% sidelap covering the area of interests in
two neighbouring strips. Z/I-Imaging provided the images with
a size of 13824 x 7680 pixels in CIR colour mode. The
georeferencing of the blocks was achieved by processing solely
DGPS and INS data with POSEO and subsequently applying an
automatic aerial triangulation with ISAT. No ground control
points were used initially. The stereo models were perfectly free
of y-parallaxes indicating an internal consistent block geometry.
The DMC is one the first operational digital aerial cameras. It
uses a modular design based on several overlapping CCD frame
cameras. It provides both panchromatic and mulitspectral
images (Hinz et al., 2001).
23.2 Laser Scanner: There were two flights with the
“Falcon” airborne laser scanner system from TopSys
Topografische Systemdaten GmbH. The first flight was in
spring after snowmelt but prior foliation. The second flight took
place in summer 2002. The average point density was 5pts/m" in
the spring flight and 10pts/m? in the summer flight. First and
last pulse data was collected during both flights. The TopoSys
System is based on two separate glas fibre arrays of 127 fibres
each. Its specific design produces a push-broom measurement
pattern on ground. For further details see Wehr and Lohr
(1999).
Sensor type Pulsed fibre scanner
Wave length 1560 nm
Pulse length 5 nsec
Scan rate 653 Hz
Pulse repetition rate 83.000 Hz
Scan with 14.3?
Data recording first and last pulse
Flight height 800 m
Size of footprint 0.8 cm
Table 2: System parameters during the Laser Scanner flight
3. METHODS
3.1 Data Preparation
3.1.1 DMC images: After the initial georeferencing just
using transformational parameters without any ground control
the blocks showed a significant absolute shift of about 50m
both in planimetry and height. Thus, a second aerial
triangulation was carried out using GCP's which were
photogrammetrically derived from another block being
available for this test site (photo scale 1:10000. c — 300 mm).
The absolute accuracy of the re-georeferenced DMC blocks
determined at several check points was found to be 20cm in
planimetry and Im in height. This is fully consistent with the
theoretical expectations for the absolute accuracy of the second
block. DSM's were subsequently calculated with ISAE which is
a software package for automatic DTM generation using image
correlation. The software takes full advantage of the 12 bit