流域水文过程的数据同化研究

发布时间：2017-12-26 16:13

本文关键词：流域水文过程的数据同化研究　出处：《武汉大学》2017年硕士论文　论文类型：学位论文

【摘要】：流域水文过程研究对于水资源管理、农业管理和自然灾害防治等方面都有着重要的影响,但应用水文模型时常常会遇到参数难以率定、预测精度差等问题。利用数据同化技术能实时融合不同类型的观测数据,从而有效提高水文模型状态和参数估计的效果。本文总结了集合卡尔曼滤波(EnKF)算法的发展概况,基于半分布式水文模型SWAT建立了水文数据同化模型,并检验了该模型的计算性能;基于数值试验讨论了径流、土壤水、地下水位等多种观测集合的影响;基于美国的Little Washita流域的实际算例研究了不同来源、不同深度的土壤水观测对水文状态的影响。本文具体的研究内容和主要结论如下:(1)建立基于SWAT模型的水文数据同化算法,并进行了计算性能的验证。针对径流过程的主要状态和参数进行数据同化,试验证明EnKF算法能够有效地降低参数和状态的不确定性,提高模型的预测效果。(2)建立可融合多源观测的数据同化模型,基于数值试验分析了径流、含水量和地下水位等观测集合对于同化效果的影响:对于径流过程而言,仅需加入径流观测就能准确地估计径流和参数CN2;浅层含水量观测可以用于估计暴雨径流,但模拟垂向水分运动时会产生偏差,不能改善整个径流过程估计;由于地下水位的变化可以在一定程度上反映土壤的垂向运动信息,补充地下水位观测可以提高土壤水的估计效果。因此,采用多源观测能更为精确地模拟径流和土壤水分状态,可以考虑使用成本低、易获取的地下水观测来反映深层土壤水的变化情况。(3)基于数值试验讨论遥感土壤水数据在水文数据同化中的价值:同化表层土壤水能用于改善上层土壤水状态估计,但是随着土壤深度的增加,同化效果逐渐削弱。(4)基于美国Little Washita流域的实例,讨论不同来源、不同深度的土壤水观测和径流观测在水文数据同化中的价值:同化径流能有效的改善流域尺度上的径流状态模拟,减小雨季时各层土壤水估计的偏差,但是在旱季的更新作用比较有限;加入表层土壤水观测能够降低下层土壤的不确定性,且对于第二层土壤的作用最明显;同化多层土壤水能从一定程度上改善径流估计的偏差。最后总结了本文研究成果,并对需要进一步完善的研究方向提出了一些建议。
[Abstract]:Hydrological process research plays an important role in water resources management, agricultural management and natural disaster prevention. However, when applying hydrological models, problems such as difficult parameter calibration and poor prediction accuracy are often encountered. The data assimilation technology can be used to fuse different types of observation data in real time, thus effectively improving the state of hydrological model and the effect of parameter estimation. This paper summarizes the collection of Calman filter (EnKF) algorithm development, hydrological data assimilation model semi distributed hydrological model based on SWAT, and test the performance of the calculation model; based on Numerical Simulation on the impact of collection of runoff, soil water and groundwater and other observations discussed the effects of soil water; observation of the actual Little Washita basin in America the examples of different sources and different depth of hydrological condition based on. The main contents and main conclusions of this paper are as follows: (1) the Hydrological Data Assimilation Algorithm Based on SWAT model is established, and the calculation performance is verified. Data assimilation is carried out for the main state and parameters of runoff process. Experiments show that EnKF algorithm can effectively reduce the uncertainty of parameters and states, and improve the prediction effect of the model. (2) establish a data assimilation model fusion of multi-source observation, numerical and experimental analysis of the runoff, water and groundwater observation set for assimilation effect based on: for the runoff process, only need to add the runoff observation can accurately estimate the parameters of runoff and CN2; shallow water observation can be used to estimate storm runoff. But the simulated vertical water movement will produce deviation, can improve the whole process of runoff estimation; because groundwater level changes can reflect the soil vertical motion information to a certain extent, of underground water level observation can improve the estimation effect of soil water. Therefore, multi-source observation can simulate runoff and soil water status more accurately, and consider the low cost and easily acquired groundwater observation to reflect the change of deep soil water. (3) based on numerical experiments, we discussed the value of remote sensing soil water data in hydrological data assimilation: assimilating surface soil water energy to improve the estimation of upper soil water status, but with the increase of soil depth, the assimilation effect was gradually weakened. (4) the Little Washita basin based on examples, discusses the value of soil water and runoff observation from different sources and different depth in hydrological data assimilation: assimilation can improve the watershed scale runoff runoff conditions on effective simulation, each layer of soil water decreased during the rainy season estimation deviation, but in the dry season update function limited; join surface soil water observation can reduce the uncertainty of the lower soil, and for the second layer soil is the most obvious effect; multi-layer soil water from assimilation to some extent improve the estimation of runoff deviation. Finally, the results of this study are summarized, and some suggestions for further improvement are put forward.
【学位授予单位】：武汉大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P333

【相似文献】