Abstract:
Leaf chlorophyll content is central to carbon, water and energy exchange between the
biosphere and the atmosphere, also to the terrestrial ecosystem function. Quantitative estimates of
leaf chlorophyll content with hyperspectral imagery can provide scientific insight for assessing
plant’s growth and stress as affected by abiotic and biotic factors. However, few studies have been
conducted the application of spectral indices in estimation of leaf chlorophyll contents of plants in
karst areas, especially in South China. After a review of the application of common spectral
indices in estimation of leaf biochemistry parameters, we found that most of the common spectral
indices were developed based on the difference, simple ratios, normalized difference and inverse
difference formulation of leaf spectral reflectance. Therefore, we firstly measured the raw
reflectance spectra of leaves from four typical karst species, namely Vitex negundo, Rhus chinensis,
Celtis sinensis and Alchornea trewioides with a ASD Field Spec 4 (Analytical Spectral Devices,
Inc., Boulder, Colorado, US) spectrometers. We then used the above-mentioned four formulations
to process the raw reflectance spectra and their first-order derivative spectra. Finally, we analyzed
the relation between leaf chlorophyll contents and relative leaf raw reflectance spectra and their
first-order derivative spectra, and tried to propose the best spectral index for estimation leaf
chlorophyll content of the plants of karst areas in South China. The results were as follows: (1)
Among the common spectral indices, the modified normalized difference vegetation index
(mND705) performed well in estimation leaf chlorophyll contents of four typical karst species in
term of the determination coefficient (R2 was equal to 0.45) and root mean squared error (RMSE
was equal to 0.26 mg•g-1). (2) However, most of the common spectral indices were not suitable for
estimation leaf chlorophyll content of the plants in karst areas. Thought the prediction capability
of fluorescence ratio index (FRI1) and chlorophyll absorption area index (CAAI) were almost the
same in estimation of leaf chlorophyll content of plants in karst and non-karst areas, their accuracy
of prediction was relative low according to the determination coefficient. (3) The spectral indices
proposed in this study performed well in estimation leaf chlorophyll content of the plants in karst
areas either based on the raw reflectance spectra or their first-order derivative spectra compared
against others common spectral indices, especially for the difference spectral index based on the
first-order derivative spectra (dD(760, 769)). Its determination coefficient was 0.71 and the root mean squared error was 0.19 mg•g-1. We, therefore proposed that the difference spectral index
based on the first-order derivative spectra (dD(760, 769)) can be used for estimation leaf
chlorophyll content of the plants in karst areas. Our results concluded that leaf chlorophyll content
of plants in karst areas can be quickly and quantitatively estimated using spectral index combined
with hyperspectral remote sensing data. These results can also provide scientific insights for
estimating plants’ growth and their adaptation to environmental stress.