Long Term Prediction of Drinking Water Demand: (Case Study of Neyshabur City, Iran)

Potable water in Iran, like most parts of the world, is scarcer every day under the effect of factors such as the drought, population growth, and increasing per capita consumption. In these conditions human effort for optimum use of this non-replacable resource seems necessary. In this regard, estimates of water demand in the future, provides the opportunity for decision makers to consideri the constraints and leading disasters to adopt of the necessary measures. In this study, the per capita water demand forecasts from both point and interval approach is used. After estimating water demand function, the possible scenarios for the future of the independent variables are predicted. Then effects of changes in economic parameters in the water use for domestic water demand are discussed. The results emphasize the importance of commodity water. In order to point prediction, four scenarios were defined and independent variables were predicted. Percent of the first three scenarios of water demand (for the case that the subsidies plan is not implemented) for the period 2011 to 2032 is 40 to 57 percent. In the fourth scenario that assumes subsidies plan, amounts of per capita water demand decreases 3 and 2 cubic meters, resepectively for years 2011 and 2013 compared to the situation that the subsidies were not implemented. Finally, by use of neural networks, the long-term water demand is predicted. The results showed that the use of indecisive part of time series in demand prediction by neural networks can solve the introspection problem of these models to some extent.