Influence of Spectral Bandwidth and Position on Chlorophyll Content Retrieval at Leaf and Canopy Levels

Abstract

Foliar chlorophyll content in vegetation ecosystems plays an important role in plant photosynthesis and it also can indicate vegetation stress and disturbance. Vegetation indices (VIs) which derived from the hyperspectral remote sensing data offer a nondestructive method for quick estimation of leaf and canopy chlorophyll content. However, the accuracy of chlorophyll retrieved by VIs depends on the spectral band information. It is a big problem how to select the suitable band and how to effectively evaluate the band position and bandwidth in chlorophyll estimation using remote sensing data. This study was focused on the impact of spectral band position and bandwidth on the chlorophyll retrieval at leaf and canopy levels. One hundred leaf spectral reflectance samples which were measured with an ASD spectroradiometer and 954 canopy reflectance samples which were simulated with the PROSAIL model were used in the study. First, the variable interval spectral average method was used to generate the plant spectra with different band position and bandwidth at leaf and canopy level. And then, linear regression model was used to build the relationship between chlorophyll (Chl) and two-band VIs (Normalized Difference Vegetation Index, NDVI; Chlorophyll Index CI; Simple Ratio Index SR). With this model, the accuracy of Chl was evaluated when varying spectral band position and bandwidth. At last, by comparing accuracies, the optimized spectral band position and bandwidth for Chl retrieval were identified. The results showed that the bandwidth and position of the spectral data could significantly affect the accuracy on Chl retrieval at both leaf and canopy levels. The optimal bandwidth on retrieval Chl would be less than 30 nm and greater than 10 nm at both leaf and canopy levels when using linear regression model. At leaf level, the optimal bands were in 748–786 nm and in 713–732 nm respectively. At canopy level, the optimal bands were in 770–750 and 765–738 nm respectively when using two-band VIs (NDVI, SR, and CI) to estimate Chl.