Estimation of Irrigation Water Demand on a Regional Scale: Combining Positive Mathematical Programming and Cluster Analysis in Model Calibration

Estimation of Irrigation Water Demand on a Regional Scale: Combining Positive Mathematical Programming and Cluster Analysis in Model Calibration

Davide Viaggi (University of Bologna, Italy) and Meri Raggi (University of Bologna, Italy)
DOI: 10.4018/978-1-61350-456-7.ch407
OnDemand PDF Download:
$37.50

Abstract

Mathematical programming tools are widely used to simulate agriculture water use thanks to their ability to provide a detailed technical and economic representation of farm choices. However, they also require a significant amount of basic information and appropriate methods for the organization of such information. The objective of the paper is to test a methodology for the estimation of irrigation water demand using a combination of Positive Mathematical Programming (PMP) at farm level, and a cluster analysis. The methodology is applied in an area of Northern Italy. The main outcome of our empirical application is the variety and complexity of reactions of different farms. The scenarios considered highlight the potential importance of the effects of price and cost variables, while the changes in the (area-based) tariff system appear less significant. The change in water cost/pricing appears somehow relevant, but does not motivate major changes in present water management policy, at least in the range of scenarios considered.
Chapter Preview
Top

The literature on simulation of irrigation system behavior provides a number of different applications including different scales and different timing for simulation. The scale of application ranges from single plants/plot models to regional scale. The timing of simulation ranges from daily water use to multiannual water use simulation.

The range of modeling approaches is also very broad, with each approach requiring different information sources and having differing capabilities with respect to decision-making processes.

An extreme example is provided by Pulido-Calvo & Gutiérrez-Estrada (2009) who use a hybrid methodology combining a Feedforward Computational Neural Network, fuzzy logic, and a genetic algorithm to forecast daily water demands one-day ahead in irrigation districts. These models were developed using historical time series data. Similarly, Pulido-Calvo et al. (2007) evaluate the performance of linear multiple regressions and Feedforward Computational Neural Networks.

Complete Chapter List

Search this Book:
Reset