2. METHODS 
The site used in this study was Black Wood, a 2.7 km 2 Forestry' Commission woodland 15km north of 
Winchester, UK. This was the principal location for a project funded by the UK National Rivers Authority, 
investigating the hydrological impacts of deciduous woodlands (Harding et al., 1992), and for this, a number of 
scaffolding towers had been erected to make meteorological measurements above the tree canopy. The two 
towers selected for the present study were located at the centre of a stand of common beech ( Fagus sylvatica) 
which were 21 m tall, with a sparse herbaceous layer, and a stand of common ash ( Fraxinus excelsior ), which 
were 16m tall, where a vigorous undergrowth of hazel ( Corylus avellana) and bramble ( Rubus sp. ) exists, with 
a dense ground cover of Dog’s mercury (Mercurialis perennis). 
2.1 Experimental procedure 
2.1.1. Spectral reflectance measurements 
Data were collected at Black Wood on one day each month for a year. A Spectron Engineering SE590 
spectroradiometer from the UK Natural Environment Research Council Equipment Pool for Field Spectroscopy 
(NERC-EPFS) served as the principal sensor for measurements made above the tree canopy . The sensor head, 
which had a nominal 15° field-of-view, was mounted on the end of a cantilevered aluminium boom and 
connected to the control unit at the bottom of the tower. A 35mm camera was also attached to the end of the 
boom, and the instruments were aligned to point vertically downwards. The camera was used to give a 
permanent record of the canopy components within the field-of-view of the SE590 at the time that each 
reflectance spectrum was obtained, and could help identify any spectra taken with the sensor head wrongly 
positioned. At the ash tower, the instrument array was supported 3 m above the canopy surface, 24m above 
ground level, while at the beech tower these distances were 5m and 26m respectively. Reflectance 
measurements were made from positions chosen to ensure that the canopy was adequately sampled, and the 
surface being measured was not shadowed by the towers. Al the ash tower, four positions were marked on the 
top bar, three at the beech tower, and at each position, two target and one reference scans were made at each of 
three distances out from the tower. 
The target spectra were normalised to percentage reflectance by dividing each by the appropriate irradiance 
spectrum, applying a calibration for the panel reflectance properties, and accounting for channel shifts which 
can occur in SE590 spectra, using software from the NERC-EPFS (Rollin, 1992). Hence, the reflectance spectra 
derived from different data collection days, under different illumination conditions, using different sensor heads 
and reference panels, were directly comparable. All the reflectance spectra from each tower, at each date, were 
averaged, giving 13 mean spectra for each tower. A filter function was applied in turn, to the mean spectra, in 
order to simulate SPOT-HRV in-band reflectances. The HRV in-band reflectances were then used to calculate 
the Normalised Difference Vegetation Index (NDV1). The position of the ’red edge’ in each reflectance 
spectrum was identified as the wavelength of the peak, between 680 and 750nm, in the corresponding 
derivative spectrum, i.e. the point of maximum slope (Horler, 1983). Curran et al.( 1991) noted that the 
calculation of derivatives eliminates additive constants (e g. illumination changes) and reduces linear functions 
(e g. linear increase in background reflectance with wavelength) to constants, which is important in the context 
of this study. 
2.1.2. Collection of canopy cover data 
Hemispherical photographs were taken looking upwards from beneath the tree canopy using a 35mm camera 
with a semi-fish-eye lens attachment. The photographs were taken early in the morning when the low sun 
zenith angle produced an even sky brightness. At each tower, four positions were permanently marked at the 
ground and the camera was placed in exactly the same locations on each occasion. For every exposure made, 
the camera was supported 1.3m above the ground, the top of the camera was aligned with magnetic north, and 
the film plane was levelled horizontally. Each photograph was analysed on an Automatic Image Analysis 
System (IBAS: Interaktiven Bild Analysen System) by placing it beneath a video camera. The video image was 
then digitised, edge enhancement and contrast stretches applied and a threshold grey level entered interactively 
to divide the image into areas of tree canopy and sky. The percentage of tree canopy within the photograph was 
then calculated by the system. 3 
3. RESULTS and DISCUSSION