GENERALISED WHOLE-FARM STOCHASTIC DYNAMIC ...

Description of the multi-period risk programming model 32Financial inputs necessary are the costs of obtaining and supplying water to the crops, irrigationsystem investment costs and salvage income, crop yield and prices, overheads and householdexpenses. The model also allows for borrowing activities and the limits on these activities mustbe specified. Interest, discount and tax rates are also specified in this section.A modelling procedure was adopted whereby the user has to specify minimum inputs for themodel. Standard DLP applications require that resource use, production income andexpenditures, etc. over the planning horizon should correspond to the time period within the lifecycle of the crop when it occurs. For example, if it is optimal to plant lucerne in year 1 and year 2the resources used in the second year of the lucerne lifecycle established in year 1 shouldcompete with the resources used in the first year of the lucerne established in year 2 and soforth. The main function of the data manipulation part is to generate data tables from the inputparameters that fulfil in the above requirement. Although this procedure is somewhatcumbersome it provides an easy way of verifying procedures compared to a model where it isdone when the matrix is generated (equation specification). Thus, the adopted modellingprocedure simplifies the model specification part.Manipulation of data is done for both the current situation (resources already committed) andother production activates. Terminal values are also calculated based on a normative approachwhich takes any cash flow streams past the planning horizon into account.The equations used to generate the matrix for the optimisation model is first specified for adeterministic model. For the stochastic version of the model additional inputs required are thecorrelation matrix and standard deviations of activity Outputs include cash flows, resource use,shadow values of binding constraints and risk trade-off curves.In the next sections the different components of the mathematical programming model arediscussed in more detail.4.1.1 DOMAIN OF THE MODELThe domain of the programming model is determined by the number of sets and the data input.In order to define the sets appropriately it is necessary to have a clear understanding of thewater allocation problem at hand. To facilitate the description of the domain of the model andalso how to input data for the model, Figure 4.2 is used to represent the water redistributionnetwork of a hypothetical water user association.