Networks of open canals are used all over the world to support agricultural activity. Globally, it is estimated that irrigation accounts for 70% of the world’s total water withdrawals. Typically, water is transported from reservoir to farms via open canals under the force of gravity while distribution of water is realized through the adjustment of gates placed along the canals. Practically, the majority of irrigation networks in the world are managed in open loop. On most farms, the amount of land that can be irrigated via the on-farm outlet is limited by the available water at the on-farm supply point. Channels are, therefore, operated conservatively with large volumes of water in the pools not to compromise productivity which may have dramatic effects for farmers. However, due to the gravity-based nature of the irrigation system, once water leaves the basin it is committed to be used and thus this conservative mode of operation leads to oversupply which can result in losses, as water spilling out at the end of the channel is no longer available for irrigation. Research has demonstrated the advantages of closed-loop water control in order to significantly improve water distribution efficiency and productivity, i.e., more agriculture product is obtained per unit of water consumed, preserving water resources for future use, while at the same time achieving a better quality service.

Field observations performed by channel operators in Northern Victoria, Australia, have shown that strong winds blowing on the water surface against the channel flow direction have an effect on the channel dynamics and correlate with a water level drop at the downstream end of the affected pool. My research focused on characterizing the wind impact on the irrigation network and its implications for automated water management. The installation on the channels of an information technology infrastructure consisting of sensors and actuators linked through a supervisory control and data acquisition communication network opened up new avenues for water monitoring and a more in depth analysis of irrigation channels dynamics. I exploited this opportunity and as first step in my approach I contacted Rubicon Systems, the industrial partner of the project, asking for data collected during field operations in major transportation channels in Northern Victoria in highly wind affected areas. The actual measurements available for my purposes were the water levels immediately upstream and downstream of each gate obtained with pressure sensors located on the gates, gate positions, wind speed and wind direction. Using the data I received, I proposed a simple second order transfer function model that was able to approximate the water level in response to the wind blowing in the opposite direction to the flow suitable for control design. I estimated the unknown parameters in the model with least-squares. I have adopted this model in a new control architecture I implemented to compensate for the wind disturbance at a local controller level. Further, expanding on previous work by my current group at the University of Melbourne, I have built a first-principle based simulation model of a channel affected by wind. I identified the values of the parameters appearing in the model by non linear least-squares, exploiting the vast dataset I was given from the company. The model constitutes a very useful tool to test controller performances prior to actual studies.