les
ure
fa
Fen
age
and
vels
its
and
ove
with
3km
1age
is of
Vays
itats
1SSES
g an
re of
Alnus
t the
jo be
The
other
are at
lense,
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
twiggy canopy over a relatively open under-storey layer.
Vegetation in the under-storey is similar to, but rather more
sparse than that of the Mixed Woodland class. The canopy is
supported on fairly evenly spaced, single stems creating a
relatively open layer between under-storey and canopy.
Sallow
The grey willow (Salix cinerea) enjoys the damp conditions of
the Fen and forms dense thickets. Because it spreads rapidly it
is necessary to keep it in check by frequent removal along the
margins of stands. It typically grows to a height of about five to
six metres with a dense, leafy canopy above multiple stems.
There is a distinct base to the canopy shading an under-storey
of sparse grasses, nettles and herbaceous plants.
Birch
The stands of Birch typically reach a height of around 20
metres and form a relatively open canopy above randomly
spaced stems. Whilst this sometimes leads to a fairly open
under-storey layer it is more common to find young hawthorns
and sallow together with other shrubs forming a distinct sub-
canopy layer. Ultimately, these stands which are dominated by
silver birch (Betula pendula) will form mixed woodland if left
unchecked.
3.1 Collection of LiDAR data
The data used for this paper was collected on 8" October 2002
as part of an ongoing monitoring programme for the fen aimed
at exploring the value of LiDAR for understanding seasonal
vegetation dynamics. First pulse, last pulse and intensity data
were recorded using an Optech ALTM 3033 device carried on-
board a Piper Chieftain Navajo aircraft. Weather conditions
were good and the aircraft completed three, overlapping flight
lines following the long axis of the site from an altitude of 1000
metres. With a scan angle of 20 degrees and a scan rate of 33
Hz this resulted in approximately one laser point per square
metre. With a narrow beam divergence the laser footprint was
in the region of 21 cm.
The ALTM was calibrated shortly before the flight using an
established survey of the runway and a hangar roofline at
Cambridge airport. The calibration data was collected to a high
level of accuracy with a total station and was tied-in to the UK
national survey, using known locations. Just over 1 km. of the
runway was covered with a 5 metre grid of control points.
Ground control for the aerial survey was provided by locating a
Novatel, 2 Hz differential GPS receiver over a known survey
point at the same airfield. With a separation of just over 30 kms.
between the ground station and the survey site the ALTM was
operating well within its required operational parameters for
generating data to an accuracy of plus or minus 15 cms. Over
five million laser points were collected and processed with the
GPS data to generate a point cloud using Applanix POSPAC
and Optech's REALM software suites. Subsequent processing
used a combination of ArcGIS and MSExcel.
At the time of the survey all of the deciduous vegetation was in
‘leaf on’ state and there was no evidence of the onset of autumn
leaf fall. Under-storey conditions were relatively dry given the
normally damp nature of the fen's soils.
4. ANALYSIS AND RESULTS
The analysis presented here relies on simple graphs and profiles
to describe the properties of the data. In the first instance these
will be used to demonstrate how first pulse data can be used to
define the outer surface of the canopy. Following on from this
combined first pulse and last pulse data will be used to
characterise the internal, vertical structure of the canopy.
4.1 First pulse observations of the canopy surface
The first stage of the analysis focussed on the use of first pulse
data to examine canopy surface properties. Experience with the
ALTM has revealed that it is highly sensitive to very small
objects in terms of triggering a first pulse response. For
example, isolated clumps of reed heads left after clearance of
reed beds can be clearly discerned in the imagery. This is
despite there being only three of four flower heads standing 1
metre above their surroundings. From this it was inferred that
first pulse data would provide an effective mechanism for
recording the canopy top at the resolution of the selected
footprint (21 cms.).
Accordingly, four areas of 100 metres by 100 metres were
selected from stands of each of the woodland classes. First
pulse data were collected for each of the cells and histograms
showing the distribution of first pulse returns binned by height
were produced. Figure 2 shows the result for Hawthorn and is
typical of all the classes.
Hawthorn 1
FP Histogram
250 + ; 120.0096
O 200! | 100.00%
Le | 80.00%
150 1
9 S | + 60.00%
p.50. : i. 40.0096
€ 504 | t 20.0096
Le fo =, im
0 | [1 * 4:3. 1. 1 I | | I 1l i 00%
S 4 5 9 49 5 O0 5 DD I
QU QD amm FFF PE PL
M ae vae ae ve oer qr qq X.
Bin
Figure 2. First pulse histogram for Hawthorn showing
under-storey and canopy layers.
All four of the histograms show a distinctive pattern with a
small peak of observations at very low level, increasing
frequency with height and one or more distinct peaks higher up
It would appear that the first, low peak corresponds with under-
storey vegetation. Relatively less energy is returned from
intermediate levels and there is a distinct peak corresponding
with the canopy top.
It is interesting to note that the form of these histograms is
similar to that of the full return waveforms reported for large
footprint sensors (Lefsky er a. op. cit.). However, there is one
important difference. In the case of these sensors energy is
being measured from throughout the canopy in response to a
single laser hit. In this case, first pulses are recorded as soon as
the laser encounters a canopy component. This means that sub-
1087
