Distributed versus Lumped Optimization of Cropping Pattern and Water Resources Utilization

Several whole-farm agro-economic optimization models have been developed to deal with lumped planning issues in the agriculture sector. However, these models cannot be used to devise appropriate management strategies at land parcel level, because of the differences between farm characteristics, and the increased complexity of the hydrological processes. Based on Spatial Farm Database (SFD) which is consisted of a number of farm-level spatial data, including location, paddock properties, owner specifications and budgets, it is possible to provide the farm manager with some suggestions regarding the optimal choice of crops and the area to be allocated for each one. To this end, genetic algorithm is used in order to cope with model nonlinearity and a large number of decision variables. In order to test the proposed model, the Mobarakabad district is modeled with 126 agriculture fields, and the optimization model is run for this area. Results showed that the optimization procedure can find more realistic farm-level optimal solutions due to its advantage in adequate modeling of field characteristics, common groundwater resources, and the associated constraints. The results of lumped optimizations could also be used as benchmarks for the purposes of comparison and interpretation.