A distributed model for urban water demand prediction

Abstract

The prediction of urban water demand is likely to play a significant role in the planning of urban water supply and drainage pipe network. The traditional water demand forecasting methods, including the water index method and the fuzzy water demand prediction model, are widely used with some deficiencies. The water demand prediction results varied among different water quantity indexes with subjectivity. The prediction results by fuzzy water demand prediction model, with fuzzy model structure, are lack of spatial distribution. Meanwhile, the process of urban water demand is closely related to the land use of urban unit. The difference of intensity and characteristic of water demand is obvious among different land use types. In this paper, a distributed urban water demand prediction model was established based on the relationship of water demand process, considering of the eight types of urban construction land use data and three types of major water demand data. In this model, we assumed the linear relationship between water demand and land use unit, ignored the fire water demand because with more randomness than other water demand process. The distributed model for urban water demand prediction was applied in Xiamen. We obtained the land use data and water demand data in 2004–2014 of Xiamen from official statistics. The monolayer construction area and the number of building stories in the residential units obtained by the high definition satellite image and the actual measurement were used to calculated the area of residential land unit, while the area of other land unit was obtained from official land planning in 2020 of Xiamen. The water demand parameters of every urban construction land unit were calibrated by PEST, and the rationality of parameters was analyzed. We observed the stability of the water demand parameters in the model while the results showed great similarity with different calibration period in the same validation period. The water demand of Xiamen and its spatial distribution in 2020 were predicted by the distributed water demand prediction model. The results showed that the water demand of Xiamen in 2020 will reach to 366.57 million tons, which will increase by 24.2% than that in 2014; the water demand intensity of residential land and industrial land will be greater than other types of construction land; the water demand intensity of residential land in Xiamen Island will be greater than the other residential land. The simulation results proved that the spatial distribution of water demand has positive correlation with population density in urban area. We observed the great difference of water demand intensity among different types land use with great difference in the process of urban water demand. The results also showed the excellent performance of the distributed model for urban water demand prediction both in calibration and validation when applied in Xiamen urban water demand prediction. Compared with the traditional urban water demand prediction model, the distributed water demand prediction model, with bright application prospect, can obtain the distributed urban water demand data with relatively stable parameters which can be determined by water demand data and land use data in the past years.