分布式水文模型SWAT的改进研究

发布时间：2018-07-01 14:17

本文选题：枯水径流 + 分季节率定　；参考：《福建师范大学》2016年博士论文

【摘要】：分布式水文模型对流域水文循环的物理、化学和生物过程都有丰富的刻画,且考虑了气象、下垫面因子的空间异质性,是认识、模拟、分析及预测变化环境下水文过程时空变化规律的有效工具。Soil and Water Assessment Tool (SWAT)模型是基于流域尺度的一个长时段的分布式流域水文模型,具有很强的物理基础,能够有效模拟和预测长期连续时间段内环境变化与管理措施对大面积复杂流域的水、沉积物和营养物输出的影响,在全世界各地得到了广泛应用。但是,SWAT模型在机理的描述、功能的提高与应用的拓展等方面还存在一些不足之处。针对这些不足,进一步完善SWAT模型,对揭示流域水文规律,提高模型模拟精度,拓展模型应用范围,提高模型应用水平,都具有重要的理论和现实意义。本文从四个方面开展SWAT模型的改进及优化工作,一是针对SWAT径流模型对枯水期径流模拟效果较差的问题,以晋江流域为例,提出将径流分为丰水序列和枯水序列分别进行率定和模拟,然后通过合并模拟“最优”的丰水序列和枯水序列组成完整径流系列的方法,并开发相应的SWAT-SC模型,实现丰枯季分开自动率定与模拟,提高枯水径流的模拟效率。二是针对加拿大地盾景观的森林流域特征,通过改进SWAT-CS模型,将氮湿沉降作为模型的驱动数据之一,以体现其动态变化对流域氮循环的影响,并修正氮湿沉降随大孔隙径流在土壤垂向分配模式,开发了SWAT-CSenm模型,以提高SWAT模型在该流域硝态氮通量的模拟和预测能力。三是利用云计算技术和GLUE方法开发了一个并行率定和不确定分析的工具—Cloud-CUT,以实现高速的SWAT模型率定和不确定分析。四是通过将分布式水文模型、WebGIS、云计算技术及其他开源软件无缝的集成,实现友好和准实时的流域管理决策系统,为流域的科学管理提供实时、简单和便捷的决策支持。研究结果表明：(1)依据丰水期和枯水期分开率定的方案构建的自动率定和模拟的SWAT-SC模型,可以有效地提高晋江流域枯水期、丰水期及全序列日径流模拟效果,对枯水期改善尤为显著。(2)针对加拿大地盾景观地区的森林流域的特点,在SWAT-CS的基础上开发的SWAT-CSenm模型,可以更加精确地刻画流域氮循环过程,有效提高模型的硝态氮模拟和预测能力。(3)利用云计算技术和GLUE方法实现SWAT模型并行率定和不确定分析的工具——Cloud-CUT,可以实现高速的SWAT模型率定和不确定分析目标。(4)通过将分布式水文模型、WebGIS、云计算技术及其他开源软件无缝集成构建的流域管理决策系统,既能满足系统功能要求,又能够满足界面直观、操作简单、决策支持实时等非功能要求,有利于促进信息共享。综上所述,通过SWAT模型的这些改进与拓展,提高了SWAT模型的模拟精度,实现了SWAT模型的高速率定和不确定的快速分析,促进了流域管理科学化水平的提升。
[Abstract]:The distributed hydrological model has a rich description of the physical, chemical and biological processes of the hydrological cycle of the watershed, and takes into account the spatial heterogeneity of meteorological and underlying surface factors. Soil and Water Assessment tool (SWAT) is a long-term distributed watershed hydrological model based on watershed scale, which has a strong physical foundation. It can effectively simulate and predict the effects of environmental changes and management measures on the water, sediment and nutrient output of large and complex watersheds over a long continuous period of time, and has been widely used all over the world. However, the SWAT model has some shortcomings in the description of mechanism, the improvement of function and the expansion of application. In view of these shortcomings, it is of great theoretical and practical significance to further improve the SWAT model for revealing the hydrological law of the watershed, improving the accuracy of the model simulation, expanding the scope of application of the model and improving the application level of the model. In this paper, the improvement and optimization of SWAT model are carried out from four aspects. Firstly, aiming at the problem of poor simulation effect of SWAT runoff model in dry season, the paper takes Jinjiang River Basin as an example. The method of dividing runoff into high water series and low water sequence is put forward, and the whole runoff series is composed by combining "optimal" abundant water sequence and low water sequence, and the corresponding SWAT-SC model is developed. In order to improve the efficiency of dry water runoff simulation, automatic rate determination and simulation are realized. Secondly, according to the forest watershed characteristics of Canadian shield landscape, through the improvement of SWAT-CS model, the wet deposition of nitrogen is taken as one of the driving data of the model, so as to reflect the influence of its dynamic change on the nitrogen cycle of the watershed. The SWAT-CSenm model was developed to improve the simulation and prediction ability of SWAT model for nitrate fluxes in the watershed. Thirdly, a parallel rate determination and uncertainty analysis tool-Cloud-CUTS is developed by using cloud computing technology and GLUE method to achieve high speed SWAT model rate determination and uncertainty analysis. Fourthly, the distributed hydrological model WebGIS-cloud computing technology and other open source software are integrated seamlessly to realize a friendly and quasi-real-time watershed management decision-making system, which provides real-time, simple and convenient decision support for watershed scientific management. The results show that: (1) the SWAT-SC model of automatic rate determination and simulation based on the scheme of separating high and low water periods can effectively improve the simulation effect of dry, high water and whole sequence daily runoff in Jinjiang River Basin. (2) according to the characteristics of forest watershed in Canadian shield landscape area, SWAT-CSenm model developed on the basis of SWAT-CS can describe the process of nitrogen cycle more accurately. (3) using cloud computing technology and GLUE method to realize parallel rate determination and uncertainty analysis of SWAT model-Cloud-Cutt, can achieve high speed SWAT model rate determination and uncertain analysis target. (3) using cloud computing technology and GLUE method to realize parallel rate determination and uncertain analysis tool of SWAT model, we can achieve high speed SWAT model rate determination and uncertain analysis target. (4) Watershed management decision system, which is constructed by integrating distributed hydrological model WebGIS, cloud computing technology and other open source software seamlessly. It can not only meet the functional requirements of the system, but also meet the non-functional requirements, such as intuitive interface, simple operation, real-time decision support and so on, which is conducive to promoting information sharing. To sum up, through these improvements and extensions of SWAT model, the simulation accuracy of SWAT model is improved, and the high rate determination and rapid analysis of uncertainty of SWAT model are realized, and the scientific level of watershed management is promoted.
【学位授予单位】：福建师范大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：P333

【参考文献】