Risk-based agricultural water allocation under multiple uncertainties

Abstract

A risk-based interval-stochastic optimization modelling approach is developed for agricultural water allocation in response to the complexity arising from uncertainties and risk in agricultural water management systems. The approach comprises conditional-value-at-risk (CVaR) model, inexact two-stage stochastic programming (ITSP) model with imprecise probabilities (IPs), and random-boundary intervals (RBIs) within a general framework. The approach can simultaneously balance expected benefits, penalties, and risks from agricultural water allocation, and can address uncertainties of agricultural water supply and demand in the form of probability distributions and intervals with random boundaries. As demonstrated in Hulan River irrigation area, northeast China, the objective of the approach is to allocate limited agricultural water resources to make a trade-off between various subareas under different risk-aversion levels and possible runoff discharges. Most inputs to the approach are expressed as interval numbers that are generated by statistical simulation, based on which various agricultural water allocation schemes are obtained. Irrigation water performance based on the optimal results are also analyzed. Results validate the applicability of the approach incorporating multiple uncertainties and risk-aversion measures in optimization models, and generating agricultural water allocation schemes in the form of interval numbers.