A bilevel modeling approach for optimizing irrigation canal scheduling under a hierarchical institutional arrangement

Well-structured institutional arrangements for irrigation canal scheduling facilitate the implementation of integrated water governance. However, there is a structural mismatch between the actual hierarchical institutional arrangements of canal scheduling and the traditional single-level canal scheduling models. Single-level models simplify hierarchically related decision-makers (HDMs) into a centralized decision-maker (CDM), which casts doubt on the feasibility of single-level models for canal scheduling in multi-level canal networks. Accordingly, we propose a bilevel canal scheduling approach involving a pair of HDMs, i.e., an agency at the upper level and farmers at the lower level. The HDMs have two assignments for making decisions on canal water allocation plans and water use plans, respectively. For comparison, two traditional models are developed: (1) one CDM with two assignments (CDM-2A), (2) one CDM with one assignment (CDM-1A), which responsible only for canal water allocation plans. These models are applied to post-harvest supplementary irrigation in Hetao Irrigation District. The results show that the planned water use in the HDM-2A model is 3.81 million m3 lower than that in the CDM-2A model, which is more in line with the water-saving willingness of farmers. Furthermore, we find that the CDM-2A model outperforms the HDM-2A model in reducing the deviation of water allocation plans from water use plans, and the CDM-1A model outperforms the HDM-2A model in reducing the deviation of water allocation plans from field water demands. However, the better performances of the CDM-2A and CDM-1A models are achieved by making farmers' water use plans exceed their real demand or depriving their decision-making rights, so the results of the two models cannot be realized in canal scheduling practice. Our findings provide new insights into the formation mechanism of the gap between theoretical canal scheduling models and practices, and also provide a reference for designing irrigation water management regimes.