Quantifying Leaf Chlorophyll Concentration of Sorghum from Hyperspectral Data Using Derivative Calculus and Machine Learning

Abstract

Leaf chlorophyll concentration (LCC) is an important indicator of plant health, vigor, physiological status, productivity, and nutrient deficiencies. Hyperspectral spectroscopy at leaf level has been widely used to estimate LCC accurately and non-destructively. This study utilized leaf-level hyperspectral data with derivative calculus and machine learning to estimate LCC of sorghum. We calculated fractional derivative (FD) orders starting from 0.2 to 2.0 with 0.2 order increments. Additionally, 43 common vegetation indices (VIs) were calculated from leaf spectral reflectance factor to make comparisons with reflectance-based data. Within the modeling pipeline, three feature selection methods were assessed: Pearson’s correlation coefficient (PCC), partial least squares based variable importance in the projection (VIP), and random forest-based mean decrease impurity (MDI). Finally, we used partial least squares regression (PLSR), random forest regression (RFR), support vector regression (SVR), and extreme learning regression (ELR) to estimate the LCC of sorghum. Results showed that: (1) increasing derivative order can show improved model performance until certain order for reflectance-based analysis; however, it is inconclusive to state that a particular order is optimal for estimating LCC of sorghum; (2) VI-based modeling outperformed derivative augmented reflectance factor-based modeling; (3) mean decrease impurity was found effective in selecting sensitive features from large feature space (reflectance-based analysis), whereas simple Pearson’s correlation coefficient worked better with smaller feature space (VI-based analysis); and (4) SVR outperformed all other models within reflectance-based analysis; alternatively, ELR with VIs from original reflectance yielded slightly better results compared to all other models.

Evaluation of Point Hyperspectral Reflectance and Multivariate Regression Models for Grapevine Water Status Estimation

“…Since PLSR extracts the shared variance between predictor variables, the higher VIP values compared to predictor variables, the less correlation is between them. This observation is similar to the findings of Bhadra et al [71]. However, they used Pearson correlation and RFR instead of Spearman correlation and RFECV to select variables.…”

Early Detection of Plant Viral Disease Using Hyperspectral Imaging and Deep Learning

“…The scoring mean was then set as a threshold to retain meaningful VIs. The MDI was selected because of its robustness in avoiding the multicollinearity (high correlation among features) [ 35 ].…”

“…To reduce spectral data to manageable low-dimensional data, the most widely used approaches are spectral band selection and data transforms. The former approach is used to choose a discrete number of key wavelengths at various positions in the spectrum to calculate representative indices (e.g., vegetation Indices) [ 21 , 35 , 36 ]. As the band selection approach preserve as much spectral information as possible, the data transform approach utilizes a transformation to compact the data into a new optimal size.…”

An Evaluation of Eight Machine Learning Regression Algorithms for Forest Aboveground Biomass Estimation from Multiple Satellite Data Products

“…Among the eight algorithms, the MARS, RF, ERT, GBRT, SGB, and CatBoost methods provided the coefficients measuring the contributions of the prediction variables in the various AGB models. The base model of the RF, ERT, GBRT, SGB, and CatBoost algorithms was the CART model, and the mean decrease in impurity importance was used to rank the importance of features [99]. In the MARS algorithm, the contribution of each predictor was determined using the generalized cross-validation statistic [100,101].…”