A distributed interval nonlinear multiobjective programming approach for optimal irrigation water management in an arid area

Abstract

Considering the uncertainties in agricultural system and spatiotemporal variability in evapotranspiration and precipitation, a distributed interval nonlinear multiobjective programming (DINMP) model was developed for optimal allocation of limited irrigation water resources in the middle reaches of Heihe River basin. The meteorological data from meteorological stations were used to estimate reference crop evapotranspiration (ET0) through FAO56 Penman-Monteith (PM) equation, and then the remote sensing MOD16/PET data were fitted by linear regression model according to the FAO56 PM results. The 95% confidence interval was used to further improve the accuracy of the fitting results. Thus, satellite-based potential evapotranspiration (PET) and ground-based ET0 estimation were integrated to not only reflect the spatial and temporal variability but also guarantee the accuracy of the ET0. In the terms of precipitation, spatial interpolation was used to spatial information of precipitation. Based on these spatiotemporal data, a DINMP with three objectives, including maximizing economic benefits and water saving as well as minimizing water shortage of critical growth periods, was formulated, and further solved by fuzzy coordination method. The optimal allocation scheme improves the irrigation water productivity by [0.50, 0.66] kg/m3, and decreases net irrigation water allocation by [0.33, 1.01] ×108 m3. These results show that DINMP can not only consider the uncertainties and multiple objectives in agricultural water management, but also improve the spatial resolution of optimal water allocation strategies. The framework of this study can provide a reference for agricultural water managers in similar areas to obtain more reasonable water allocation schemes.