Exploring the optimum spectral bands and pre-treatments for chlorophyll assessment in sunflower leaves from yellowness index

The Yellowness Index (YI) was originally developed for evaluating manganese deficient soybean leaves, but it has been successfully applied to indicate chlorosis in stressed leaves of other plant species. Despite distinct vegetal species present very similar spectral signatures, there are subtle differences in their reflectance patterns and magnitudes that influence the performances and the wavelengths used to calculate spectral indices. In this study, an algorithm was developed, capable of finding the best wavelengths for assessing chlorosis of leaves using the YI. The proposed algorithm was tested with spectral reflectance measurements for estimating the chlorophyll content of sunflower (Helianthus annuus L.) leaves submitted to different water stress levels. Original spectral signatures were pre-treated by centring, normalization and detrending methods prior chemometric analyses and results were also evaluated. Both original and modified YI resulted in suitable predictions of sunflower leaf chlorophyll content. The modified YI based on spectra pre-treated with detrend method, and centred between 662 and 750 nm with a band separation of 44 nm, reached higher r2 value (82.31%) and lower RMSE (2.28 and 2.67 µg cm−2) both for calibration and validation datasets, when compared with the results of the other tested pre-treatments and spectral ranges. The proposed algorithm was efficient to search better performances for YI, finding the best wavelengths for assessing chlorophyll content of sunflower leaves. It can be easily used for different chlorosis reference measurements and plant species. When implemented in a software package the proposed algorithm resulted in an effective tool, quickly performing thousand tests by using files containing many spectra and sample data.