Mathematical control theory tools are not yet widely applied to agricultural systems, although they could be used to improve the optimisation of crop management and serve as a basis for decision support tools. In particular, modern crop models are generally computer models with complex mathematical structures that limit the ability to apply efficient optimization techniques. On the other hand, decision models do not necessarily need to be as detailed as models for deep understanding of internal processes and furthermore, limited decision variables and relatively poor online measurements also advocate for simple models. In this context, the development and use of control oriented models will be presented with applications towards the reuse of wastewater for irrigation. First, it will be shown how reduced models can be used with soil water content data to deal with sensor faults and spatial variability in soil structure. Secondly, a double modelling method is proposed that allows to use in parallel a detailed crop model and a reduced order model for the resolution of optimal control problems in crop management.