Automation for Water and Nitrogen Deficit Stress Detection in Soilless Tomato Crops Based on Spectral Indices

Abstract

Water and nitrogen deficit stress are some of the most important growth limiting factors in crop production. Several methods have been used to quantify the impact of water and nitrogen deficit stress on plant physiology. However, by performing machine learning with hyperspectral sensor data, crop physiology management systems are integrated into real artificial intelligence systems, providing richer recommendations and insights into implementing appropriate irrigation and environment control management strategies. In this study, the Classification Tree model was used to group complex hyperspectral datasets in order to provide remote visual results about plant water and nitrogen deficit stress. Soilless tomato crops are grown under varying water and nitrogen regimes. The model that we developed was trained using 75% of the total sample dataset, while the rest (25%) of the data were used to validate the model. The results showed that the combination of MSAVI, mrNDVI, and PRI had the potential to determine water and nitrogen deficit stress with 89.6% and 91.4% classification accuracy values for the training and testing samples, respectively. The results of the current study are promising for developing control strategies for sustainable greenhouse production.