canner
ber of
it gaps
eflects
ible to
i] laser
ial tree
e laser
locate
stems,
boreal
y data
y. The
pecies
ample
ectare
utting
taking
were
wn in
'. The
anced
on by
1) was
orded.
rating
^a and
or (the
on, the
object.
Along
angle,
Juha Hyyppä
and capability to obtain the profile information requires careful validation. High measurement density is required in
order to be able to detect individual tree crowns. Steep incidence angle enables to have sufficient number of ground hits.
Test flights (TopoSys, 1996) have shown that at incidence angles of more than 10° off-nadir, the amount of shadowed
areas heavily increases, i.e., the number of measured ground hits decreases and gaps in the DTM occur more frequently.
The profiling capability is typically limited in laser scanners to few modes. Typically both the first and the last pulse are
included; referring to the first and last echoes of distributed targets, such as the forest.
There are a small number of the airborne laser scanners available on the market today. Main providers are TopoSys,
Optech and Saab Survey Systems. TopoSys laser scanner was selected for the study due to its high measurement density
and steep incidence angle.
The laser scanner campaign was carried out on 2-3 September 1998. TopoSys-1 laser scanner was installed in the local
aircraft. Three DGPS receivers were employed to record the carrying platform position: one on board the aircraft, and
two ground reference GPS stations (the first as basic receiver, the second for backup). The 2-by-0-5-km test site was
intensively flown from the altitude of 400 m resulting in measurement density equivalent of more than 10 measurements
per m°. The survey altitude was selected in order to guarantee the number of pulses needed to separate individual trees.
Due to the survey altitude applied, the swath width was approximately 100 m. Both the first and last pulse modes were
collected, but only first pulse mode was used for the creation of the 3-dimensional height model of the forest.
3 PRE-PROCESSING OF LASER DATA
Laser scanner survey provided a cloud of points, the x, y and z coordinates of which are known. They form a digital
surface model (DSM), which includes terrain points, vegetation points, and points reflected from buildings. By
processing the data and classifying the points to terrain and vegetation points, it was possible to produce digital terrain
model (DTM) and digital vegetation model (DVM). When only the top of the vegetation is included in the model, it can
be called digital crown model (DCM). The difference between the DCM and DTM models is called in this study a
digital tree height model (DTHM), 3-dimensional representation of the tree heights within the target forest area.
3.1 DTM Generation
There exist several algorithms to produce the DTM, but a rather new approach (Pyysalo, 2000; Hyyppi et al., 2000c) is
described here in brief.
The generation of DTM includes five phases (Pyysalo, 2000):
Calculation of the original reference surface
Classification of vegetation and removal of vegetation from the reference surface
Classification of the original cloud of points using the reference surface
Calculation of the DTM based on classified ground hits
Interpolation of missing points
Ut Bio be
Calculation of the original reference surface was done by transforming the cloud of x, y and z co-ordinates into a grid
and by recording the minimum terrain height z of all points corresponding to certain cell location. Classification of
vegetation was performed using filtering. It was assumed that the ground surface is continuous and terrain elevations do
not significantly change locally. A gradient of the matrix was calculated as a sum of differences of near-by pixels and as
a sum of absolute values of differences of near-by pixels according to the formulas. The calculated gradient values were
compared with a threshold value: if either of the gradient values was less than the threshold, the elevation corresponding
to pixel was labelled as ground hit. Otherwise it was assumed as vegetation hit. The new elevations for the vegetation hits
were calculated by Delaunay interpolation algorithm and by using of the terrain heights of near-by pixels. The process is
iterative, typically 3 to 4 iterations are recommended. Classification of the original cloud of points was performed using
the calculated reference surface. The terrain height of original points was compared against the calculated reference
surface and the difference dz, was calculated as follows
dz, = z, — 2(j,i) (1)
where EL the terrain height of original points, and
z(ji) is the terrain height of corresponding pixel in calculated reference surface.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 423