Presenting a Next Generation Irrigation (NGenIrr) demand model

Abstract

In this paper we describe a rural water demand (irrigation) model that captures the behavioural complexities and uncertainties associated with current irrigated farming realities. This is done through stochastic compromise programming and crop-water simulations, integrating hydrologic, biophysical and behavioural factors associated with irrigation water demands. Behavioural factors are modelled by finding the compromise between two conflicting objectives (maximize gross margins and minimize risk of crop water-stress). The weightings on these objectives are input as a discrete distribution to represent varying degrees of irrigators’ risk-averseness. Input data and modelling errors (and uncertainty) are incorporated by introducing a multiplicative error on the crop water simulation outputs. The output from the model is a probability distribution of water demand. The model can be used at various spatial and temporal scales (from paddock(s) to irrigation district(s) and from daily to monthly). These features of the model are the basis for calling it a next generation irrigation (NGenIrr) demand model. The model has been developed using eWater CRC’s The Invisible Modelling Environment (TIME) software development framework. The application of NGenIrr demand model to the Shepparton Irrigation District is also discussed in this paper. The model has been calibrated and validated using observed water demand data over 10 irrigation seasons. Although the model achieved a good calibration (r2=0.70), the validation results (r2=0.54) suggest that the model needs to be improved further. This research work is part of larger eWater CRC projects investigating uncertainty analysis in models and improving water management decisions.