Building a metamodel of an irrigation district distributed-parameter model

Complex decision-making problems, related to planning and management of irrigation water resources, generally preclude the use of large, distributed-parameter models, which are then commonly substituted by lumped-parameter models. This paper, with the aim of improving the quality of these latter, introduces a new approach for their design. This approach is based on metamodelling, which proposes to identify a simple, lumped-parameter model on the basis of the data produced via simulation with a distributed-parameter model. The approach proposed is tested on a real-world case study, namely the identification of a metamodel describing the water demand of the Muzza–Bassa Lodigiana irrigation district (Italy). The metamodel, which inherits the physical description of the original distributed-parameter model, is sufficiently simple to permit the resolution of an optimal control problem, i.e. the design, via stochastic dynamic programming, of the release policy of Lake Como, serving the Muzza irrigation district.