Raw Data, Processing Code, PLSR Model, Correction Factors, and Clean Data of Hyperspectral Heather Measurements and Expert Classification of RGB Images

We developed an experimental procedure for hyperspectral measurements of heather plants and classification of hyperspectral data using a PLSR (Partial Least Squares Regression) model. Here we provide our full data set including hyperspectral signatures of healthy, stressed, and dead heathers (Calluna vulgaris, variety: Sanne) measured from the cutting to the young plant stage (26.03.2019 - 18.06.2019). The visual classification of each heather plant at each measurement day was performed by two experts, the position and vitality status of each plant is given in the excel files (Expert Classification.xlsx). Results from our root test approach are given in the pdf file (Root_Test_Results.pdf). The hyperspectral data includes raw data (Raw Data.zip), calculated correction factors (Correction_factors.txt) and the processed data (Clean Data.zip). In addition, we provide the developed R-code for application of the PLSR model and the code for data processing with all necessary functions (Calluna_Code.zip), which includes the correction of the spatial heterogenity of the illumination conditions using a white reference sheet that was spectrally characterized by spectral laboratory measurements to calculate correction coefficients. The data processing procedure to compensate for the spatial heterogenity of the illumination conditions and defined regions of interest per plant as well as a step-by-step explanation of the code and PLSR model results are explained in our manuscript: Hyperspectral imaging for high-throughput vitality monitoring in ornamental plant production, authored by Ruett M, Junker-Frohn LV, Siegmann B, Jaenicke H, Whitney C, Luedeling E, Tiede-Arlt P, and Rascher U. We do not provide support for further data processing.