基于WRF模式和HEC-HMS水文模型的西苕溪流域洪水预报研究

发布时间：2018-09-06 18:33

【摘要】：洪水预报精度和预见期作为洪水预报的两个关键要素,与防洪减灾效益密切相关,洪水预报越精确,预见期越长,防洪调度的经济价值就越大。为了延长洪水预见期,保证洪水预报精度,其中一个比较有效的方法就是在洪水预报中耦合高分辨率天气预报模式。WRF模式凭其便捷和完善的模拟与预报系统,在中尺度模拟特别是降水模拟中具有普适性和优越性,但将其应用于耦合洪水预报的研究还较少。西苕溪流域是我国东部典型的中尺度流域,位于长江中下游地区,其气候特点和地形地貌特征在中国东南部半湿润地区具有典型性和代表性。因此,本文以西苕溪流域为研究区,采用港口站以上流域的实测降水和港口断面的实测洪水资料,开展了基于WRF模式和HEC-HMS水文模型的洪水预报研究,实现了气象模式与水文模型的单向耦合,并由此建立了一套适用于中尺度流域的实时洪水预报方案。研究包括以下几个部分： 1.根据西苕溪流域的气候水文特征,从2009～2011年间4场洪水过程资料中筛选出两类典型的暴雨洪水过程,分别为2009年8月的台风型暴雨洪水和2011年6月的梅雨型暴雨洪水,利用NCEP再分析资料,分析其发生天气背景和雨洪特性。在此基础上,采用WRF模式分别对这两个暴雨个例进行了连续10天的滚动模拟,并对模拟结果进行了检验和分析。结果表明：WRF模式对主要雨日降水落区、走向、雨强的模拟结果与实测吻合度较高。对子流域单元面雨量进行了时空分级检验,发现WRF模式对强降雨的预报能力较强,当出现lmm以下小雨时(日面雨量lmm),模式容易发生漏报。时间上,随着预报时效的延长,单元预报合格率虽有所下降,但仍维持在50%左右；空间上,子流域面积越大,地形起伏越大,则面雨量预报误差越大。总体来看,WRF模式对研究区两类暴雨过程的模拟结果较为合理和可靠,可作为水文模型的输入开展耦合预报试验。 2.基于地理信息处理和模型参数估算结果,采用两种产汇流方案组合,对西苕溪流域进行了降雨—径流过程模拟,并以《水文情报预报规范》(GB-22482-2008-T)中的作业预报许可误差作为合格检验标准分析了模拟效果。结果表明,模拟日径流和次洪过程与实测一致较好,日径流模拟结果中,确定性系数、洪峰流量相对误差、峰现时差等指标合格率为100%；次洪模拟结果中,仅有第20110611次洪水(组合二)的确定性系数为不合格,说明HEC-HMS模型在西苕溪流域具有较好的适用性。模型对于单峰型洪水的模拟效果优于双峰型洪水,其参数组合更适用于短历时的降雨事件,组合一较组合二有更好的表现,说明在资料相对缺乏的情况下,相对简单的参数方案更具有实用价值。 3.采用单向耦合法,将WRF模式(5km网格)48h预见期的滚动预报降雨场,经尺度转换,输入IEC-HMS模型进行流量滚动预报,成功地实时预报出了该流域的两类洪水过程。并且随着时间的推移,滚动预报精度逐渐提高,洪峰和洪量得到不断调整,当降雨过程趋于结束时,滚动预报结果也趋于稳定。耦合预报相对于传统预报方法显示出一定的优势,使洪水预见期有效延长48h,且具有一定的预报精度。上述三部分研究表明：WRF模式降雨预报误差是制约流量预报的关键因素,洪峰流量和峰现时间的误差取决于WRF模式对降雨落区、强降水时段预报的偏差。
[Abstract]:As the two key factors of flood forecasting, flood forecasting precision and forecasting period are closely related to the benefit of flood control and disaster reduction. The more accurate flood forecasting, the longer forecasting period, the greater the economic value of flood control operation. WRF model is a typical mesoscale basin in the eastern part of China, located in the middle and lower reaches of the Yangtze River, with its convenient and perfect simulation and forecasting system. It has universality and superiority in mesoscale simulation, especially in precipitation simulation, but its application in coupled flood forecasting is less. Climatic and topographic features are typical and representative in the semi-humid area of Southeast China. Therefore, this paper takes Xiweixi River Basin as the research area, uses the measured precipitation above the harbor station and the measured flood data of the harbor section to carry out the flood forecasting research based on WRF model and HEC-HMS hydrological model, and realizes meteorology. A set of real-time flood forecasting schemes suitable for mesoscale basins is established based on the unidirectional coupling of the model and hydrological model.
1. According to the climatological and hydrological characteristics of Xiweixi River basin, two typical storm flood processes were selected from the data of four flood processes during 2009-2011, namely, typhoon-type storm flood in August 2009 and meiyu-type storm flood in June 2011. The weather background and rainstorm flood characteristics were analyzed by NCEP reanalysis data. WRF model was used to simulate the two rainstorms for 10 days, and the simulation results were checked and analyzed. The results show that the WRF model has a high coincidence with the measured rainfall intensity in the main rainy sunset area. The prediction ability of the model for heavy rainfall is strong, and it is easy to miss the forecast when there is light rain below LMM (daily surface rainfall lmm). Generally speaking, the simulation results of WRF model for two types of rainstorm processes in the study area are reasonable and reliable, and it can be used as the input of hydrological model to carry out coupled forecasting experiments.
2. Based on the results of geographic information processing and model parameter estimation, the rainfall-runoff process in Xituoxi watershed was simulated by using two runoff generation and confluence schemes, and the simulation results were analyzed by using the operational forecast allowable error in (GB-22482-2008-T) as the eligible test standard. In the simulation results of daily runoff, the qualified rate of deterministic coefficient, relative error of peak discharge and current time difference is 100%; in the simulation results of secondary flood, only the deterministic coefficient of 20110611 flood (combination 2) is unqualified, which indicates that the HEC-HMS model has a good applicability in Xituoxi River basin. The simulation effect of single-peak flood is better than that of double-peak flood, and the parameter combination is more suitable for short-duration rainfall events. The performance of combination one is better than combination two, which shows that the relatively simple parameter scheme has more practical value in the case of relative lack of data.
3. Using one-way coupling method, the rolling forecast rainfall field of WRF model (5km grid) in 48h forecast period is transformed into IEC-HMS model by scale transformation, and the two types of flood processes are successfully predicted in real-time. With the passage of time, the rolling forecast precision is gradually improved, and the flood peak and flood volume are constantly adjusted. The results of rolling forecast also tend to be stable when the rainfall process tends to end. Compared with the traditional forecast method, the coupled forecast method shows certain advantages, which can effectively prolong the flood forecasting period by 48 hours and has a certain forecasting accuracy.
The above three parts show that the error of WRF rainfall forecast is the key factor to restrict the discharge forecast, and the error of peak discharge and peak present time depends on the deviation of WRF model to rainfall area and heavy rainfall period forecast.
【学位授予单位】：南京信息工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P338

【参考文献】