School of Environment, Tsinghua University, Beijing, China; Department of Industrial Engineering and Operations Research, University of California, Berkeley, CA, USA.
School of Environment, Tsinghua University, Beijing, China.
J Environ Manage. 2018 Oct 1;223:658-667. doi: 10.1016/j.jenvman.2018.06.073. Epub 2018 Jul 1.
Urbanization, one of the predominant trends of the 21st century, places great stress on urban water supply networks. This paper aimed to identify the most important variables driving urban water supply patterns in China, a region which has seen rapid urban growth in the last few decades. In addition, a principal component analysis-informed urban water sustainability index was developed in order to benchmark cities. The research involved applying statistical learning and other analytical methods to 12 years of urban water supply data for 627 cities across China. The findings were as follows: (1) PCA showed that approximately 46.8% of variability in the data could be explained by two principal components. Component 1 (37.26%) was more closely associated with variables related to water supply and sale, supply pipelines, and water supply finance. C2 (9.51%) was clearly related to urban water prices and average per capita water use. (2) Random forest and XGBoost algorithms were effective in classifying cities according to their region, with model testing accuracies of 87.69% and 88.32% respectively. (3) Chinese cities have consistently suffered water loss/leakage rates above 20% since 2001, and water prices are closely associated with leakage. (4) China's urban water sustainability has increased by just 3.56% between 2001 and 2013; Southwest China saw the highest growth rate in urban water supply sustainability. The implications of our research effort will be useful for decision makers in water-stressed urban areas around the world who are seeking novel insights in how to leverage statistical learning techniques to gain insights into urban drinking water supply patterns.
城市化是 21 世纪的主要趋势之一，给城市供水网络带来了巨大压力。本文旨在确定推动中国城市供水模式的最重要变量，中国在过去几十年经历了快速的城市化增长。此外，还开发了基于主成分分析的城市水可持续性指数，以便对城市进行基准测试。该研究涉及应用统计学习和其他分析方法，对中国 627 个城市的 12 年城市供水数据进行分析。研究结果如下：（1）主成分分析（PCA）表明，数据的大约 46.8%的可变性可以用两个主成分来解释。第一主成分（37.26%）与供水和销售、供水管线和供水财务有关的变量密切相关。第二主成分（9.51%）与城市水价和人均用水量明显相关。（2）随机森林和 XGBoost 算法有效地根据城市的地理位置对城市进行分类，模型测试准确率分别为 87.69%和 88.32%。（3）自 2001 年以来，中国城市的水损失/泄漏率一直持续在 20%以上，水价与泄漏密切相关。（4）2001 年至 2013 年间，中国城市水的可持续性仅增长了 3.56%；西南地区城市供水可持续性增长最快。我们的研究成果将对全球水资源短缺城市的决策者有用，他们正在寻求利用统计学习技术洞察城市饮用水供应模式的新见解。
