用水需求的时空演化规律及其分布式模拟预测

Water demand analysis and modeling are the foundations for water resource planning, engineering projects design and water resource management. In existing water demand studies, spatial-temporal evolutions of water demand and its driving factors are carried on a fixed time-space scale, the relationship between water demand and driving factors is assumed to be stationary, and lumped models are often taken for simulation and forecasting of water demand on the basis of the statistics or investigation unit scales. However, these pre-requirements and lumped models are not always satisfied for the significant nonstionary evolution and highly spatial heterogeneity of water demand under the climate change and intensified human activities. Our proposal will mainly focus on the three aspects of water demand. The variation of water demand at different spatial-temporal scales should be first investigated through statistical methods, and then the scale transfer functions for water demand will be figure out. The sizes of spatial and temporal scales for study area with water supply projects under the changing environment will be determined through the investigation of the functions, operation rules of the projects. A distributed water demand model for simulation and forecasting will also be developed with the methods of system dynamics (SD) and cellular automata (CA). Our study will reveal the mechanism of water demand at multi-scales. And the results from our study will not only enhance the knowledge of the water demand, but also improve the accuracy of water demand forecasting, which will support the delicacy water resources allocation, the strictest water resources management system and the adaptive water resources management under the changing environment in China.

用水需求分析和模拟预测是进行水资源规划、工程设计和水资源管理的基础。现有的用水需求研究常以固定时空尺度来分析用水需求及其驱动要素的变化规律、假定驱动关系符合平稳性要求以及采用以统计调查单元为尺度的集总式模拟预测等，而这些条件和手段现难以反映出气候变化和人类活动影响下用水需求呈现出的多尺度耦合、动态性演化和异质性分布等特点。本项目将围绕用水需求在不同时空尺度上的变化规律，重点研究用水需求、驱动要素和驱动关系各自的时空尺度转换问题，辨识变化条件下用水需求适应性时空尺度的大小，并构建用水需求的分布式模拟预测模型。以期建立多时空尺度的用水需求理论与方法，从而揭示用水需求的驱动机理，增强用水需求过程及其空间分布规律的认知，有效解决水资源规划和配置中存在的来、用水时空尺度不匹配问题，提高用水需求预测精准性，从而为水资源精细化配置和最严格水资源管理制度的实施等重大国家需求提供理论基础与技术方法支撑。

用水需求分析和模拟预测是进行水资源规划、工程设计和水资源管理的基础。现有的用水需求研究常以固定时空尺度来分析用水需求及其驱动要素的变化规律、假定驱动关系符合平稳性要求以及采用以统计调查单元为尺度的集总式模拟预测等，而这些条件和手段现难以反映出气候变化和人类活动影响下用水需求呈现出的多尺度耦合、动态性演化和异质性分布等特点。本项目以我国地级市生活、工业、农业用水数据为基础，围绕用水需求在不同时空尺度上的变化规律，重点研究了：用水需求的时空分布特性与演化规律、用水需求的适应性时空尺度判别、用水需求的分布式模拟预测三个方面的内容。主要结果如下：（1）我国地级市的历史用水数据表现出显著的非平稳性与较弱的空间自相关性。生活用水需求的驱动要素为城市生活、第三产业、农村生活、牲畜的用水量，人口与第三产业增加值总额；工业用水需求的驱动要素为纺织业、造纸业、冶金业等11个工业行业的增加值总额及其相应用水量；农业用水需求的驱动要素为潜在蒸散发、归一化植被指数、降水与土壤含水量。（2）经济效应是影响生活与工业用水需求的重要因素，而农业用水需求则主要取决于潜在蒸散量；通过对比不同时空尺度下驱动要素对用水需求的累计贡献，识别出生活、工业、农业用水需求的适宜时间尺度与空间尺度，分别为季、月、旬与1km、1km、500m。（3）结合用水需求的适应性时空尺度，构建用水需求分布式模拟预测模型，模拟结果与地级市尺度的历史用水数据高度吻合，表明该模型模拟效果良好。