Leaf chlorophyll is vital for plants because it helps them get energy through the process of photosynthesis. The present review thus examines various leaf chlorophyll content estimation techniques in ...laboratories and outdoor field conditions. The review consists of two sections: (1) destructive and (2) nondestructive methods for chlorophyll estimation. Through this review, we could find that Arnon's spectrophotometry method is the most popular and simplest method for the estimation of leaf chlorophyll under laboratory conditions. While android-based applications and portable equipment for the quantification of chlorophyll content are useful for onsite utilities. The algorithm used in these applications and equipment is trained for specific plants rather than being generalized across all plants. In the case of hyperspectral remote sensing, more than 42 hyperspectral indices were observed for chlorophyll estimations, and among these red-edge-based indices were found to be more appropriate. This review recommends that hyperspectral indices such as the three-band hyperspectral vegetation index, Chlgreen, Triangular Greenness Index, Wavelength Difference Index, and Normalized Difference Chlorophyll are generic and can be used for chlorophyll estimations of various plants. It was also observed that Artificial Intelligence (AI) and Machine Learning (ML)-based algorithms such as Random Forest, Support Vector Machine, and Artificial Neural Network regressions are the most suited and widely applied algorithms for chlorophyll estimation using the above hyperspectral data. It was also concluded that comparative studies are required in order to understand the advantages and disadvantages of reflectance-based vegetation indices and chlorophyll fluorescence imaging methods for chlorophyll estimations to comprehend their efficiency.
Valeriana jatamansi Jones is an aromatic herb well known for its essential oil contents, and its high medicinal and commercial values. The amount of essential oils present in it increases with ...maturity (age) of the plant. In this study, Hyperspectral remote sensing data recorded in the field using Analytical Spectral Devices (ASD) handheld spectroradiometer was used to discriminate the age (6, 12, 24 and 36 months) of V. jatamansi. Principal Component Analysis (PCA) was used for feature selection and 06 machine learning classifiers were used to classify the plant based on their ages, i.e., Artificial Neural Network (ANN), Support Vector Machine (SVM), Random Forest (RF), Boosting Decision Tree (BDT), Decision Tree (DT) and k-Nearest Neighbourhood (kNN). For comparison, these classifiers were applied on full range of spectral reflectance data without feature selection and on feature-selected data using PCA. It was found that the accuracies of ANN, RF, BDT, SVM, DT and kNN were 91, 85, 57, 78, 35 and 42%, respectively for non-feature selected datasets. The accuracies of ANN and DT classifiers were, respectively, increased by 100% and 75% after applying PCA. The ANN classifiers resulted in 100% overall accuracy with a Kappa coefficient (K) of 1. The wavelength regions 860, 870 to 874, 876 to 885 nm in near-infrared (NIR), and 747 to 756 nm (red-edge) were identified as regions suitable for discriminate the age groups of V. jatamansi. The final trained model thus prepared was again validated on 60 plants (with different age group) grown in its natural habitat and the obtained accuracy was 88% (K = 0.84). Thus, the present study have provided a rapid technology for onsite identification of age of V. jatamansi in the field itself. The developed technology thus provides a scientific way for harvesting of this plant at its optimum age avoiding its wastage. The results of this study can also be applied to other endangered and valuable plants by way of finding its optimum growth stages for its harvesting.
Spectral data are now-a-days widely used for assessment of biochemical, biophysical and structural traits of vegetation by analysing their spectral signatures. In field, it is recorded by handheld ...spectroradiometer sensor. In doing so, the spectral data acquisition is influenced by several factors such as variations in light intensity during recording; number of spectral readings per plant; distance between sensor and plant; impact of heating due to sun, wind, and wetness of plants. Thestandard operating practices for such data acquisition are generally based on the similar work earlier carried out by researchers in bits and pieces. In the present study experimental set-ups have been laid for systematic studies to answer the influence of above factors on spectral data and required optimisations have been suggested. It was found that variations in light intensity influence the spectral readings, when the illumination difference was more than ∼20%. The 30 spectral readings were found optimum. The reflectance spectra recorded at distance of 20-35 cm were treated as pure spectra in case of 25 °FOV sensors. The heating of samples due to sun, speed and direction of wind, and wetness of samples influenced the plant spectral reflectance . It was also observed that averaging of spectra recorded in several observations (instead of single observation) for the same plant samples optimize the errors. It was concluded that in order to have a good quality of vegetation field spectral data, case to case calibrations, and care must be taken to eliminate influence of surrounding environmental variables while spectral data acquisition to avoid/minimize the errors.
The Indian Himalaya region harbours approximately 1748 plants, which have been prioritized for medicinal usage. To ease the pressure on these plants, the government of India is encouraging the ...in-situ cultivation of medicinal plants. As a consequence, Saussurea costus, Valeriana jatamansi, and Picrorhiza kurroa are some of the important crops which are being cultivated on large scale owing to their high market demand, conservation value and medicinal properties. Identification of these plants in the field requires taxonomic skills, which is one of the major bottlenecks in the conservation and management of these plants. In this background, a hyperspectral library of the above three medicinal plants has been prepared by collecting its spectral data from Himachal Pradesh and Uttarakhand states of Indian Himalaya. The Random forest (RF) model was implied on the spectral data for the classification of these medicinal plants which resulted in training accuracy of 84.39 % (kappa coefficient = 0.72) and testing accuracy of 85.29% (kappa coefficient = 0.77). This RF classifier has identified Green (555-598 nm), red (605 nm), and NIR (725-840 nm) wavelength regions suitable for discrimination of the above species.