Development of multi-hazard risk assessment model for agricultural water supply and distribution systems using bayesian network

To identify and assess the impact of various hazards threaten agriculture water supply systems, the development of a risk analysis framework is inevitable in promoting sustainable agricultural development. This study aims to develop a novel multi-hazard risk assessment model by applying a Hybrid Bayesian Network for agricultural water supply and distribution systems. This model consists of discrete and continuous nodes and their probable interactions. The structure of this model is designed to assess the risk associated with the agricultural water system's for supply and distribution sections separately considering various factors such as river discharge, the inflow of water distribution system and its fluctuation, and the demand of water. The developed model is applied to the Roodasht Irrigation district located in the center of Iran to demonestrate its capability. This Irrigation district is under the threat of drought, improper performance of the ditch-riders, and operational losses. The results showed that the model in both training and test datasets has proper accuracy and performance with the root mean square error of 0.069 and 0.076, coefficient of determinations equal to 0.717 and 0.690, and the overall index of model performance equal to 0.787 and 0.671, respectively. The results of this research and the proposed model will help stakeholders and decision-makers to be aware of the probable causes and the extent of system failure and its components due to the threatening hazards. Also, it will assist in planning the allocation of irrigation water based on predictable risk associated with various hazards.