Journal Article

Agricultural Water Allocation by Integration of Hydro-Economic Modeling with Bayesian Networks and Random Forest Approaches

Roozbahani, Abbas, Sherafatpour, Zohreh, Hasani, Yousef

2019

Overview

Sustainable utilization of water resources requires preventive measures that must be taken to promote optimal use of water resources together with consideration of stakeholder interests and the economic value of water. The main objective of this study is to present an integrated hydro-economic model for allocating agricultural water based on its economic value. The study region covered six irrigation networks downstream of the Zayandeh Rood Dam in Iran. In fact, this study addresses questions of how to allocate scarce water to different consumers, in order to achieve the highest efficiency and economic benefits. To gain this goal, the existing agricultural activities in each irrigation network were simulated by applying the Positive Mathematical Programming (PMP) economic model and then by coupling the economic model with a water allocation planning model of the basin (MODSIM), the hydro-economic framework was generated. These tools helped to allocate water based on its economic value, maximize net profit by determining the optimal cultivating area and analyze the effects of various allocation scenarios on employment, economic productivity, and reliability indicators. Moreover, Bayesian Networks and Random Forest algorithms were developed as an automated substitute to simplify the process and reduce computational complexity. The results showed that the Nekoabad Network enjoys top priority followed by the Barkhar, Mahyar, Sonati, Abshar, and Rodasht Networks. After implementing the Bayesian Network, the four criteria of MAE, MAPE, R2, and the Nash-Sutcliffe index for the irrigation networks were 9 MCM, 24%, 0.738, and 0.644 respectively, which indicated the model has good accuracy. Random Forest method was also employed as a novel technique in automated allocation, and the values obtained for the four mentioned criteria were 7 MCM, 15%, 0.861, and 0.80, which showed it is more accurate. The results indicated the capability of the presented hydro-economic model as well as the intelligent models substituting it in allocating agricultural water.

• Publisher: Springer Nature B.V
• Subject: Agronomy / Algorithms / Automation / Bayesian analysis / Computer applications / Computer simulation / Econometric models / Economic analysis / Economic benefits / Economic conditions / Economic models / Economics / Forest management / Irrigation / Mathematical models / Mathematical programming / Model accuracy / Network reliability / Networks / Probability theory / Water allocation / Water resources / Water scarcity / Water use