241
computing the reflectance. This gave rise to the apparent reflectance of the plots from the helicopter altitude
(Figure 1). To compute the NDVI the reflectance in broad red and near infrared wavebands was simulated using
the spectral response functions of the Daedalus AADS 1268 Airborne Thematic Mapper. The red-edge position
was calculated using the method of Guyot et al. (1992) which involved linear interpolation between reflectance
in narrow red and near infrared wavebands.
4.3 Main results
After reviewing the video track, two plots were excluded from the analysis because of poor positional accuracy.
Data analysis concentrated on the relationships between plot LAI and the high spectral resolution reflectance
data, the NDVI and the red-edge position.
4.3.1 High-spectral resolution reflectance. A correlation spectrum between plot LAI and the reflectance in each
of the Spectron channels was computed (Figure 2). There were statistically significant correlation with LAI in
the 500-660nm region and around 700nm but no significant correlations were found in the 750-800nm region
of the near infrared.
4.3.2 Correlation with NDVI. A positive correlation between plot LAI and the NDVI was observed although
the relationship was strongly affected by the data for the plot with the lowest LAI. There appeared to be no
relationship for plots with a LAI greater than 6 .
4.3.3 Correlation with red-edge position. The experiment showed that, the red-edge position was strongly
correlated with plot LAI. The correlation was higher than that with the NDVI (Figure 3). This result may have
implications for work on spectral shifts associated with forest decline; the red-edge position is clearly controlled
by leaf amount, as measured by LAI, as well as by the chlorophyll content of the foliage.
Figure 1. Reflectance spectrum for plot Al Figure 2. Reflectance correlation spectrum with LAI, 95%
confidence limits are shown at +/- 0.53
Figure 3. Relationship between LAI and NDVI Figure 4. Relationship between LAI and red-edge position
