Retrieval Of Vegetation Carotenoid Content Using Spectral Indices Based On Radiation Transfer Models

Being one of the most basic chemical reactions of vegetation, photosynthesis closely relates to material cycle and energy transformation, providing essential conditions for the growth of vegetation. Carotenoid, as a kind of important photosynthetic pigment in leaves, plays an important role in conveying light energy, protecting chlorophyll from over exposure and delaying the rapid decomposition of chlorophyll in senescing leaves in order to maintain the photosynthetic capacity. The accurate retrieval of carotenoid content is essential to monitor the health of vegetation. Remote sensing technology provides a convenient approach to monitor large-scale vegetation rapidly and dynamically. Moreover, the rapid development of hyper-spectral imaging technology enhances the studies on the retrieval of biochemical compositions with the use of distinctive spectral features. Retrieving carotenoid by spectral information is more difficult than chlorophyll because these two pigments have overlapped absorption features and the content of chlorophyll is much more than that of carotenoid. Hence, researches on carotenoid are far fewer than those on chlorophyll. In this paper, a large number of vegetation reflectance spectra over 400-2500nm were simulated by PROSPECT model and PROSAIL model with a wide variety of leaf biochemical and biophysical parameters. A novel spectral index for retrieving carotenoid at both leaf and canopy scales was developed by collaborating multiple spectral indices.Aiming to determine the best spectral bands of narrow band normalized difference vegetation index (NDVInb) for estimating carotenoid contents in leaf, all possibilities of combining NDVInb were tested by comparing the correlation with carotenoid contents. According to the analysis of anti-interference ability to other leaf biochemical constituents, no NDVInb is found that can be applied in the retrieval of carotenoid.At leaf level, the construction of the new vegetation index for estimating carotenoid content is concerned with the method of multi-index form. For eliminating the interferences of other factors, different mathematical forms, such as subtraction form, ratio form and NDVI form, were compared to found an index that is merely sensitive to carotenoid. Local sensitivity analysis and global sensitivity analysis (EFAST) were both taken to evaluate the new vegetation index, and finally this index passed the validation.The new index found at leaf level was applied to analyze its feasibility and inversion effect at canopy level. Through the sensitivity analyses, the inversion research of this index at canopy level was modified to improve the inversion precision at canopy level.