International Archives of the Photogrammetry, Remote Sensing 
By contrast small footprint devices have a narrow divergence 
angle resulting in footprints typically between 10 cms. and ! 
metre. However, these systems operate at very high frequency 
and may generate anything up to 80,000 laser pulses per 
second. Bandwidth constraints mean that these systems only 
record a small part of the energy for each pulse. Typically the 
height of the first detectable energy return (first pulse) and the 
last detectable return (last pulse) is measured together with the 
overall light intensity. These devices are capable of generating 
extremely dense patterns of observations over the scenes being 
imaged recording up to 20 points per square metre. By virtue 
of their remarkable accuracy they have found widespread 
application in surveying, forestry and other environmental 
applications. 
Given these basic differences in functionality it is not surprising 
that with respect to vegetation applications large footprint 
systems have been mainly used for examining the vertical 
structure of forest canopies (e.g. Lefsky ef al. (1999), Harding 
et al. (2001), vertical patterns of photosynthetically active 
radiation (e.g. Parker et al, 2001), and above ground biomass 
(Means et al. 1999). Small footprint systems on the other hand 
have been used primarily for forest survey and mensuration 
(e.g. Naesset, 2002, Persson ef al. 2002). Generally, a 
modelling framework is used in which laser height observations 
are linked to ground based measurements of key forest 
parameters (e.g. height, numbers of stems, crown basal area) 
using regression techniques. . 
To date very few researchers seem to have considered the 
possibilities of using small footprint LiDAR for monitoring the 
vertical structure of canopies. Notable exceptions include Blair 
and Hofton (1999) who demonstrated that by integrating,small 
footprint returns across an area equivalent in size to a large 
footprint it was possible to elucidate the vertical structure of 
complex rainforest canopies in Costa Rica. Riano et al. (2003) 
also suggested this possibility in the context of forest mapping 
for fire models. 
2. AIMS AND OBJECTIVES 
Given this background the present study aims to highlight the 
possibilities of using small footprint, airborne LiDAR to 
understand the structure of complex, semi-natural vegetation 
stands. 
In particular it will: 
e demonstrate how small footprint lidar can be used to 
characterise the vertical structure of dense, highly 
complex canopies. 
e draw attention to its potential for mapping gaps, 
canopy openness, density and under-storey properties 
e examine structural differences between woodland 
vegetation types and demonstrate the potential that 
LiDAR measurement of structure has for 
discrimination in classification exercises. 
3. STUDY AREA AND SURVEY DATA 
The study site for this work was Woodwalton Fen, an area of 
semi-natural wetland located just to the south of Peterborough 
in lowland East Anglia, UK (Figure 1). The Fen is one of 
and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
Britain's oldest nature reserves and has variously been 
designated as a RAMSAR site, a Special Area of Conservation 
(cSAC) and a site of Special Scientific Interest (SSSI). It 
occupies an area of approximately 3km by 2 km bounded by 
  
  
  
[ e 5 | 
| \ | 
| | 
Y z^ 
T - | y "EP d 
irl Sug | 
< L4 dS { | 
7 D | NM | 
La" | JF, Y 
: | (Coq Norwich — V 
XX i { e i 
{if Peterborough > 
La 3 el 
| 4 S9 
x Y 3 ac^ | 
| )OA yn ) | 
| JL S PT ra / 
fr | J M Cambridge x » | 
Woodwalton EN e i ol 
Fen i - j^ J 
AJ í t s A 
A Arr otim 
| AL ^ uer eet Y | 
| Ly + 
t An 
  
  
  
  
Figure 1. Location of the Woodwalton Fen study 
site 
canals from the surrounding arable farmland. It is rich in its 
variety of fauna and flora with some 47 red data book species 
and two very rare plants. The site is managed by English Nature 
who plan to link it to its neighbour, Holme Fen as part of a 
novel and ambitious conservation scheme called ‘the Great Fen 
Project’. 
Internally, the fen is divided by a network of smaller drainage 
channels which split the area into a grid of grass and woodland 
habitats. A system of sluices enables control of the water levels 
and maintenance of the damp conditions needed for its 
conservation. Peat in the surrounding farmland has dried and 
become eroded leaving the fen perched several metres above 
the surrounding landscape. The entire site is virtually flat with 
a change in elevation of approximately 1 metre across the 3km 
of its length. 
A network of grass walkways follow the internal drainage 
channels and provides pedestrian access to the different areas of 
the fen. In general cells of the grid defined by the waterways 
contain distinct vegetation habitats. It is the woodland habitats 
which were of interest to this study and four particular classes 
were examined: 
Mixed Woodland 
This class represents the tallest of the canopies reaching an 
average height of over twenty metres. It consists of a mixture of 
birch (Betula spp), Hawthorn (Crataegus spp.) and Alder (Alnus 
spp.). The top of the canopy is dominated by Birch whilst the 
Alder appears in isolated clumps. The Hawthorn tends to be 
lower, forming a sub-storey beneath the main canopy. The 
under-storey layer consists of grasses, nettles and other 
herbaceous plants emerging from a layer of litter. 
Hawthorn 
Pure stands of Hawthorn (Crataegus spp.) are relatively rare at 
the fen. However, where it is found in stands it forms a dense, 
1086 
Inte 
  
twig 
Veg 
spar 
supr 
relat 
Salle 
The 
the ! 
is ne 
marg 
six r 
Ther 
of sp 
Bircl 
The 
metre 
space 
under 
and s 
cano 
silver 
unche 
3.1 Ci 
The d 
as par 
at exp 
vegeta 
Were 1 
board 
were § 
lines f 
metres 
Hz thi 
metre. 
in the 1 
The À 
establi: 
Cambr 
level o 
nationa 
runway 
Ground 
Novate 
point at 
between 
operatir 
generati 
five mil 
GPS da 
and Opi 
used a c 
At the ti 
"leaf on 
leaf fall. 
normally