A Management Analysis Tool for Emancipated and Public Irrigation Areas Using Neural Networks

The management transfer of irrigation districts from the public to the private sector became a broad strategy throughout the world. Although the extent to which the process is being implemented, there is little information available on the results of these transfer programs. In addition, there is no established procedure to analyze economic feasibility and general performance of the irrigation districts. In this context, the aim of this research was to develop a model to evaluate the performance of self-managed irrigated areas transferred from public sector to private irrigator associations. A list of performance indicators proposed by the Brazilian Federal Court of Accounts to monitor the public perimeters and pre-classification information from two public companies, San Francisco and Parnaiba Valleys Development Company (CODEVASF) and National Department of Works Against Droughts (DNOCS) were used in this research. A statistical multivariate model with discriminant analysis (MDA) was performed to identify the indicators importance in order to discriminate the current level of the irrigation areas. The data resulting from multivariate discriminant analysis was used to create an artificial neural network (ANN) that classifies the irrigated areas related to management. It was observed that the indicator Generated Revenue per Hectare (GRH) was the most important in the discriminating process regarding self-management. The neural network created from the values of the performance function resulted from multivariate discriminant analysis showed be capable of assessing the performance of Irrigated Perimeters over time and also be adequate as a tool for resource allocation and evaluation of self-managed irrigated areas.