基于陆气耦合的降水径流预报研究

发布时间：2018-06-28 14:24

本文选题：陆气耦合 + WRF　；参考：《清华大学》2015年博士论文

【摘要】：径流预报的预见期和可靠性是水库调度及流域防洪、供水、发电等计划编制的重要前提,对提高水资源综合管理能力具有重要意义。陆气耦合模式利用降水预报信息驱动水文模型获得未来径流过程,是目前提高径流预报预见期的研究热点和前沿问题,属多学科交叉范畴。本文从数值天气模式、分布式水文模型以及陆气耦合模式的研究进展入手,从理论方法和应用检验两方面开展研究,取得了一些有意义的进展。论文首先分析了数值天气模式参数化方案优化组合的必要性,构建了用于评价模式降水预报能力的指标体系,并基于Euclid贴近度的概念,建立了数值天气模式参数化方案定量综合评价模型,使多指标评价下的参数化方案选择过程更加客观。然后,充分利用GLUE和PSO各自的优点,通过PSO算法改进GLUE的抽样规则,利用GLUE框架及PSO多次寻优和可加性判定准则改进等效参数组的获取方式,提出了基于GLUE和PSO融合的不确定性参数率定算法,提高了等效参数组的代表性,有效改善了不确定性参数率定的搜索效率。之后,以雅砻江流域为应用研究区,建立了雅砻江流域数值天气预报模式,评价得出了流域WRF模式最优参数化方案组合;并基于HEC-HMS,建立了雅砻江流域分布式水文模型,实现了与流域数值天气预报模式的单向耦合。在此基础上,利用GLUE-PSO不确定性参数率定算法构建了模型参数库(参数率定)并对其进行了验证,提出了基于模型参数库的不确定性洪水预报方法。最后使用JAVA语言,在高性能计算环境下,开发建立了基于陆气耦合的雅砻江流域洪水预报原型系统,实现了耦合模式的自动化运行和模型参数库的动态维护。应用表明,论文构建的陆气耦合模式能够显著延长径流预报预见期,可将研究区——雅砻江流域各水文站的径流预报预见期(最长约54h)提高至6d左右,提出的基于模型参数库的不确定性洪水预报方法能够充分反映预报过程中的不确定性信息,使96h以后的降水预报仍然具有较高的使用价值。研究成果对进一步结合预报信息的水库调度方法探索,以及雅砻江数字流域建设及实践应用,均有积极的意义。
[Abstract]:The prediction period and reliability of runoff forecasting is an important prerequisite for reservoir operation, flood control, water supply, power generation and so on. It is of great significance to improve the ability of integrated management of water resources. The land-atmosphere coupling model uses precipitation forecast information to drive the hydrological model to obtain the future runoff process. It is a hot topic and a frontier problem in improving the forecast period of runoff forecasting. It belongs to the interdisciplinary category. This paper starts with the research progress of numerical weather model, distributed hydrological model and land-atmosphere coupling model, and makes some significant progress in theoretical method and application test. In this paper, the necessity of optimizing and combining the parameterized schemes of numerical weather model is analyzed, and an index system is constructed to evaluate the forecast ability of the model precipitation, and based on the concept of Euclid's closeness degree, A quantitative comprehensive evaluation model for parameterized schemes of numerical weather models is established, which makes the selection process of parameterized schemes more objective under multi-index evaluation. Then, taking full advantage of the advantages of GLUE and PSO, the sampling rules of GLUE are improved by PSO algorithm, and the method of obtaining equivalent parameter groups is improved by using glue framework and the criteria of multiple optimization and additivity of PSO. An uncertain parameter rate determination algorithm based on glue and PSO fusion is proposed to improve the representativeness of equivalent parameter set and the search efficiency of uncertainty parameter rate determination. Then, the numerical weather forecast model of Yalong River basin is established, the optimal parameterized scheme combination of WRF model is obtained, and the distributed hydrological model of Yalong River basin is established based on HEC-HMS. Unidirectional coupling with watershed numerical weather forecast model is realized. On this basis, the model parameter library (parameter rate determination) is constructed by using GLUE-PSO uncertainty parameter rate determination algorithm and verified, and an uncertain flood forecasting method based on model parameter library is proposed. Finally, a prototype system for flood forecasting in Yalong River basin based on land and atmosphere coupling is developed under the high performance computing environment with Java language, which realizes the automatic operation of the coupling model and the dynamic maintenance of the model parameter library. The application shows that the land-atmosphere coupling model constructed in this paper can significantly prolong the forecast period of runoff forecast, and can increase the forecast period of runoff forecast (longest about 54 hours) to about 6 days for hydrologic stations in the studied area-Yalong River basin. The uncertain flood forecasting method based on the model parameter library can fully reflect the uncertain information in the forecast process, so that the precipitation forecast after 96 hours is still of high practical value. The research results are of positive significance to the further exploration of reservoir operation method combined with forecast information, as well as to the construction and practical application of Yalong River digital watershed.
【学位授予单位】：清华大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TV121.1

【相似文献】