231
CORRELATION ANALYSIS BETWEEN INDICES OF TREE LEAF SPECTRAL
REFLECTANCE AND CHLOROPHYLL CONTENT
Jan-Chang Chen a , Chaur-Tzuhn Chen b ’ *
a Dept, of Recreation Management, Shih Chien University. 200 University Rd., Neimen, Kaohsiung 84550, Taiwan,
China
b Dept, of Forestry, National Pingtung University of Science and Technology. 1 Shuehfu Rd., Neipu, Pingtung 912,
Taiwan, China. cct@gisfore.npust.edu.tw
Commission VII, WG-VII-3
KEY WORDS: Hyperspectral Remote Sensing, Forestry, Spectral Indices, Simulation, Correlation
ABSTRACT:
This study was conducted to investigate variations of leaf chlorophyll content and surface spectral reflectance of different tree
species {Daphniphyllum glaucescens ' Michelia formosana ' Illicium dunnianum and Machilus kusanoi) across contrasting terrain
in the Nanjenshan Reserve of Renting National Park, Southern Taiwan. The results showed that there were significant differences (P
< 0.01) among the REP ( Red Edge Position ) with total chlorophyll concentration (R 2 =0.508, Daphniphyllum glaucescen;
R 2 =0.667, Michelia compressor R 2 =0.503, Illicium dunnianum; R“=0.774, Machilus kusanoi). The vegetation index from
hyperspectral measured positively correlated with total chlorophyll concentration in all species SR705 (R 2 =0.236, P<0.000);
mSR705 (R 2 =0.5283, P<0.000); ND705 (R 2 =0.265, PO.OOO) and mND705 (R 2 =0.573, PO.000). The results suggest that the REP
could be used to estimate the chlorophyll content in tree leaves. The index mNDVI705 seemed more sensitive to detecting
chlorophyll content in a wide range of tree species across a terrain. Results obtained in this paper discuss using hyperspectral data to
estimate the actual leaf chlorophyll content by using low costing and time-saving procedures.
1. INTRODUCTION
Green plants all have unique spectral features, mainly because
of the chlorophyll and carotenoid and other pigments and water
content can together constitute the spectral feature of a plant
(Philip and Shirly, 1978). Traditionally, the predominant
method for measuring chlorophyll content is using the
spectrophotometer; with this method plant tissues are used,
which costs more time and sustain some damage to the plants.
Therefore, a portable hyperspectral instrument for testing is an
alternative method that can be carried out without hindering
chlorophyll measurements through reflection, transmittivity and
absorption (Chappelle et al., 1992). This method showed the
unique application of hyperspectral technology in estimating
chlorophyll content of vegetation. The application of
hyperspectral can help us get much information, and the
information can not be acquired from traditional wide
wavelength spectra or ratio parameters. Therefore using high
analytical spectrum instrument (e.g., AVRIS, CASI, LI-1800,
GER1500 ) allows us to directly decide the red edge reflection
value (the reflection value on the first derivative peak). In
addition, a linear method or a polynomial method may be
adopted to estimate the chlorophyll content of plants. The chief
method adopted by many predecessors, when using
hyperspectral remote sensed data to assess vegetation, is
regression method, which is used to decide the wavelengths that
are related with the content of chlorophyll. Regression analysis
(or the structure of wavelengths, including vegetation index and
differential spectra) is then conducted between these
wavelengths and chlorophyll content to establish some
regression models. Researchers have also used the above
theoretical basis to estimate the chlorophyll contents, nitrogen
nutrition condition, growing trend, effective radiation in
photosynthesis, and biomass of plants (Dalezios et al., 2001;
Thenkabail et al., 2000) and on the aspect of ecological system
there are studies on the vertical layers of forests (Blackburn,
1998). The spectral reflection feature of a leaf is usually
directly related with the surface characteristic, leaf structure,
water content and chlorophyll content (Gausman, 1982). In
order to compare the spectral difference among leaves of
different species with different shapes of leaf, in this research,
mNDVI given by Sims and Gamon (2002) was taken as NDVI
for comparison. Most studies related with (REP) are limited in
the areas of the growth of agriculture products and the
estimation of their yields. However, since few studies are
concerned with forests, the focus of this study is on the
relationship between hyperspectral data such as vegetation
index, and REP optical coefficient, and chlorophyll content
based on 4 tree species taken from 3 topographical areas such as
windward, valley, and leeward. This data can provide more
reliable information on the application of hyperspectral data in
the study of forest vegetations’ indices.
2. MATERIALS AND METHODS
2.1 Study Area
The leaf samples of four species used in this study were
collected from Nanjenshan Nature Reserve of Renting National
* Corresponding author.