Dynamic modelling of lettuce transpiration for water status monitoring

Abstract

Real-time information on the plant water status is an important prerequisite for the precision irrigation management of crops. The plant transpiration has been shown to provide a good indication of its water status. In this paper, a novel plant water status monitoring framework based on the transpiration dynamics of greenhouse grown lettuce plants is presented. Experimental results indicated that lettuce plants experiencing adequate water supply transpired at a higher rate compared to plants experiencing a shortage in water supply. A data-driven model for predicting the transpiration dynamics of the plants was developed using a system identification approach. Results indicated that a second order discrete-time transfer function model with incoming radiation, vapour pressure deficit, and leaf area index as inputs sufficiently explained the dynamics with an average coefficient of determination of RT2=0.93±0.04. The parameters of the model were updated online and then applied in predicting the transpiration dynamics of the plants in real-time. The model predicted dynamics closely matched the measured values when the plants were in a predefined water status state. The reverse was the case when there was a significant change in the water status state. The information contained in the model residuals (measured transpiration – model predicted transpiration) was then exploited as a means of inferring the plant water status. This framework provides a simple and intuitive means of monitoring the plant water status in real-time while achieving a sensitivity similar to that of stomatal conductance measurements. It can be applied in regulating the water deficit of greenhouse grown crops, with specific advantages over other available techniques.