Real-time control approaches for site-specific irrigation and fertigation optimisation

Automated irrigation and fertigation site-specific control systems offer labour and water savings, and crop productivity improvements for growers, where spatial variability of water requirements exists within a field. Real-time irrigation control strategies for surface and pressurised irrigation systems have been developed that adapt to infield soil and plant measurements collected in real-time. 'Sensor-based' control strategies directly use measurements to make irrigation decisions; and 'model-based' control strategies use a model (often calibrated with sensor input) to aid irrigation decisions. Model-based control strategies can aim for specific end of season characteristics. However, model-based control strategies often use off-the-shelf, black box industry models that may not be updated with the development of the new varieties, and may not consider all the soil-plant-water relations. A hybrid Artificial Neural Network (ANN) and Bayesian model is being used for training and predicting crop dynamics based on historical and real-time infield data. A game theory and artificial intelligence-based approach will provide an inbuilt self-learning capability for new crop conditions to achieve site-specific irrigation optimisation and real-time adaptive control. This paper will present an overview and comparison the adaptive control approaches for irrigation and fertigation, and considerations for their use under commercial conditions on surface and pressurised irrigation systems.