Résumé
Crop models allow simulating irrigated plant dynamics at the plot level. However, in many places irrigation is managed collectively to share water at the network level. To study the impact of the irrigation network constraints on plant dynamics, we proposed a co-simulation approach based on the coupling of the Optirrig crop model at the plot level with the WatASit agent-based model at the network level. As a proof of concept applied on a typical gravity network of the South-East of France, the approach allowed to consider the effects of the network spatial (i.e. water flow gradient) and temporal (i.e. network coordination) constraints on leaf area index and water stress index dynamics of 16 cereal plots. Four progressive levels of collective irrigation constraints are simulated: no collective constraints, space collective constraints, time collective constraints, and space and time collective constraints. Retrospective simulation of the 2017 irrigation campaign is consistent with field surveys, and simulation results suggest that plant water stress could be underestimated when simulated at the plot level rather than at the network level. Spatially, the most severe water stress was observed for the plants located furthest downstream of the network. Temporally, the absence of network coordination can lead to earlier plant water stress and lower plant growth during the collective irrigation campaign, while time-slot-based coordination tends to delay the impact. For future research, reinforcing the coupling from the crop model to the agent-based model could allow to study the feedback loop of plant dynamics on irrigation practice adaptations. It is also a first step towards an optimization approach for irrigation networks.