Improvement in Linearity Between Hyperspectral Vegetation Indices and Chlorophyll Content, Leaf Area Index Based on Radiative Transfer Models
基于辐射传输模型的高光谱植被指数与叶绿素浓度及叶面积指数的线性关系改进

Hyperspectral vegetation indices, which have fine spectra, can be used to detect the subtle physiological and environmental changes in vegetation. Therefore, use of the indices opens up an avenue for precise measurement for agriculture, especially for chlorophyll levels and leaf area index. However, a number of vegetation indices are not linearly related to biochemical components, so these indices may not be appropriate for other vegetative areas. We selected two different scale parameters, leaf chlorophyll content and leaf area index, on the basis of the radiative transfer models PROSPECT and SAIL, respectively, for sensitivity study of a set of indices (NDVI, MSR and MCARI). We aimed to improve the linearity between the indices for chlorophyll content and leaf area index on the basis of combinations of different wave bands. The sensitivity study of chlorophyll content revealed that for all the three kinds of indices, replacing the reflectance of 800 nm/700 nm and 670 nm with 750 nm and 705 nm, respectively can successfully improve the linearity limits of all vegetation indices. A nearly linear relationship is achieved between the index of MCARI705 and the chlorophyll content. Sensitivity study of leaf area index revealed that the index of MCARI705 can better alleviate the saturation problem of leaf area index changes. On comparing other indices, an obvious saturated region emerges for MCARI705 only when leaf area index exceeds 8. In general, different waveband selection (replacing 800 nm/700 nm and 670 nm with 750 nm and 705 nm, respectively) does not guarantee improvement of the linearity of indices with increased leaf area index as compared with chlorophyll content because of the complicated structure of the canopy.