通过分析影响我国用水数据的因素, 模拟我国各地级行政区的用水量数据, 对于在全国开展用水情况评价、水 资源高效利用与管理、实现经济可持续发展有重要意义。以 2000- 2010 年全国地级行政区内土地利用分布、人口、 经济和用水量统计数据为基础, 采用两层前馈神经网络模型模拟用水量数据。研究结果表明, 土地利用分布、人口 和经济三类影响因素综合模拟用水量具有较高模拟精度。考虑到近年来经济技术、节水措施和用水效率不断提高 对用水量的影响, 而这种影响只体现在时间上, 提出了/ 四维0 用水量模拟理念, 即在模拟用水量时补充土地利用、人 口和经济的时间序列特征。结果分析表明,/ 四维0 模拟理念进一步提高了全国各地级行政区的用水量模拟精度。

It is o f g reat sig nificance of a nationwide water use ev aluatio n, efficient use of w ater resour ces and manag ement to achiev e sustainable eco nomic dev elo pment by analy zing the facto rs affecting the w ater consumptio n data and simulatio n data in each administrative reg ion o f China. Based on the statist ical data such as land use distributio n, po pulation, eco nomy and w ater co nsumpt ion in administr ativ e reg ions of China f rom 2000 to 2010, this paper simulates the data of w ater consumption w ith a two2lay er feed2fo rw ard neura l net wo rk model. The results showed that the compr ehensive simulation o f water consumptio n for three influencing factors ( land use distributio n, po pulation and econom y) had a high simulatio n accur acy . Considering the impact of economy and techno lo gy , wat er2saving measures and incr eased water eff iciency o n w ater consumptio n in r ecent years, t his study pro posed " four2dimensional" water consumption simulation concept, which was to supplement the time series char acteristics of land use, populatio n and eco nomy. The results showed that "four2dimensio nal" simulation concept further impro ved the simulatio n accuracy o f wat er consumptio n.

国家重点研发计划( 2016YFC0401308; 2018YFC0406502) ; 国家自然科学基金( 51779007; 41671018)