Prediction of Cucumber Leaf Stomatal Conductance through Machine Learning by Using Leaf Temperature and Environmental Conditions as Input Data

In order to achieving the optimal plant growth, the suitable growth environment is necessary to provide to the plant. The ideal strategy is to provide a suitable growth environment based on plant physiological conditions. Stomatal conductance is one of the important index for the plant physiology. However, the stomatal conductance measurement is labor and time consuming. In this work, a method to predict the cucumber leaf stomatal conductance was proposed. The cucumbers (Cucumis sativus L.) were planted under well-watered and water stress conditions. During the planting period, the environmental information such as ambient temperature, relativehumidity, soil moisture, and sunlight was collected. In addition, an infrared thermal camera and a leaf porometer were used to measure the temperature and stomatal conductance of cucumber leaves, respectively. The results revealed that the leaf temperature and stomatal conductance of cucumbers under water stress were higher and lower, respectively, than those of well-watered cucumbers. The stomatal conductance of leaves under water stress was low because of the need to minimize water loss, and the leaf temperature was high because heat could not be abstracted via the stomata. Furthermore, machine learning was employed for establishing and verifying the stomatal conductance prediction model from the collected information. For the machine learning prediction results, the importance of the leaf temperature is higher than other input data. The results revealed that machine learning could be used for predict the cucumber leaf stomatal conductance.