Application of spectral indices and reflectance spectrum on leaf nitrogen content analysis derived from hyperspectral LiDAR data

Abstract

Hyperspectral LiDAR (HSL) can be applied in several fields, especially in nitrogen analysis because of its ability of obtaining more abundant spectral information. Besides, spectral indices (SIs) related to leaf nitrogen content (LNC) of vegetation have also been applied in hyperspectral remote sensing widely. Thus, this paper is mainly to use SIs derived from HSL data to analyze rice LNC. The partial least squares regression (PLSR) is utilized to link the SIs to rice LNC. By combining several more wavelengths than other SIs, two integral indices derived from HSL data, including the normalized area over reflectance curve (NAOC) and reflectance integral index (RII), have better performance in terms of LNC retrieval. Meanwhile, HSL-based reflectance spectra in more wavelengths than SIs have been processed by a method of band selection, and then been used to validate the detection efficiency and flexibility of HSL system. By selecting bands, the LNC retrieval have been optimized and improved significantly (R2 > 0.7) than using SIs. It can be seen that HSL will be a great potential and flexible technology in remote sensing because of owning more spectral information about the objects.