The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008 
234 
discrepancy caused by different leaf structure is eliminated. 
This result also shows that using the modified indices can also 
raise the correlation as mentioned above even in the calculation 
of spectra reflectance for a single species of tree leaves (the 
results of mixed species will describe in the following section). 
There was an obvious correlation between chlorophyll content 
and mSR and mNDVI (Fig. 3), which showed that these two 
indices can be used to effectively estimate the change of 
pigment content of a single species. From this result and on the 
basis of the correlation coefficient for the same species, 
mNDVI was always higher than NDVI, and there was a huge 
difference between these two sets of indices. From the analysis 
of the relationship between the pigment content and SR and 
mSR, we found that SR was in a relatively random and irregular 
relationship with the change of pigment contents; their 
correlation was different for variance species. 
Daphniphyllutn glaucescens 
Michelia formosana 
X Q.SRn ti 
"&.mND 7(ps 
YlNESs 
\x“SR>j*5 
^O-SRjds 
'O. ND '?0S 
1000 2000 3000 4000 5000 6000 7000 8000 
1000 2000 3000 4000 5000 6000 7000 8000 
Total Chlorophyll (p/g) 
Illicium dunnianum 
Total Chlorophyll (p/g) 
Machilus kusanoi 
1000 2000 3000 4000 5000 6000 7000 8000 
~ x '\mSR 7 o5 
''O,SR 70 j 
- mND ?( i 5 
\3 ND 7( i 5 
1000 2000 3000 4000 5000 6000 7000 8000 
N O.SRt* s 
' mND 7( , s 
"■nr®™ 
Total Chlorophyll (p/g) 
Total Chlorophyll (p/g) 
Figure 3. The correlation between chlorophyll content and NDVI, mNDVI, SR, mSR in 4 species 
We also analyzed the correlation of the vegetation indices and 
chlorophyll content of the samples taken from 4 different 
species of trees and the results are shown in Fig. 4 where the 
modified indices were in a better correlation with each other 
for different species. Sims and Gamon (2002) also 
demonstrated that modified spectral indices can be applied 
across species with widely varying leaf structure without the 
necessity for extensive calibration for each species. The results 
for the most useful in ecological remote sensing studies, the 
above relationships could be generalized across most species 
and different leaf ages. Modified indecies could be used for the 
leaf structural variability was simply to understand that the 
indices can remove the difference only by one major 
wavelengths 445 nm. Many studies indicated that light 
reflected directly from the leaf surface never enters the leaf 
cells and thus is not influenced by pigment and water content. 
Surface reflection can be greatly enhanced by leaf epidermis 
(e.g., leaf hairs, surface waxes, cuticle thickness, et cetera), and 
wavelengths 445 nm can detect the variation (Reicosky and 
Hanover, 1978; Slaton et al., 2001). They also demonstrated 
that modified spectral indices can be applied across species 
with widely varying leaf structure without the necessity for 
extensive calibration for each species. From the above results 
we concluded that for the correlation of mNDVI and mSR with 
the change of pigment contents of a single species, mNDVI 
was in a better correlation, and there was a huge difference 
between mNDVI and mSR in this respect. Following the 
results, the modified indices were added as a constant to all 
reflectance values reduces both on SR index and ND index 
even when there was no change in the leaf absorptance of 
tissues below the epidermis or in the same tree species.