Use of artificial neural networks in canal irrigation management

Artificial neural networks (ANNs) are becoming widely used in water resources applications. In this research a canal flow prediction system was developed using ANNs. The resulting model has been applied to the Piute canal in the Sevier River Basin of south-central Utah. The model predicted the quantity of water that must be diverted in order to satisfy future irrigation demands, while minimizing the volume of spill at the end of the canal. A sensitivity analysis was performed on the final structure of the ANN model to test the robustness of the model and to study the influence of ANN parameters. The application of the model on the Piute canal demonstrates that ANNs can be used to construct accurate models for forecasting canal flows and managing irrigation water delivery systems in near real-time.