Sensing crop reflectance for water stress detection in greenhouses
Water deficit stress is one of the most important growth limiting factors in crop production. Several methods have been used to detect and evaluate the effects of water deficit on plants. In the present study, remote sensing has been used. The objectives were to 1) determine the effects of water deficit on greenhouse crop using spectral indices and 2) evaluate the reflectance spectra using the classification tree method for distinguishing water deficit levels based on substrate water content. For this reason, tomato plants were grown in slabs filled with perlite: i) under no irrigation for five days and ii) continuously well-watered. A hyperspectral camera was used to remotely measure plant reflectance during the periods with normal or low substrate water content. A large number of vegetation and water indices were compiled in order to measure plant vigour and other biophysical parameters using the remotely sensed data. A significant correlation was found between the mrNDVI reflectance index and substrate water content (θ). In addition, the PRI values observed for plants of the water deficit treatment varied, once θ% rapidly decreased to 4.5%. The daily mean TCARI was linearly correlated with chlorophyll content but not with θ% variations. A classification tree was developed to investigate the relationship between the categorical data and determine the variables affecting other independent variables. The decision tree methodology could be used to build the model for the prediction of irrigation event outcomes.
Katsoulas, N., Elvanidi, A., Bartzanas, T., Ferentinos, K.P. and Kittas, C. (2018). Sensing crop reflectance for water stress detection in greenhouses. Acta Hortic. 1197, 117-126
remote sensing, hyperspectral, machine vision, classification tree, reflectance index