黄河洪水预报.doc

黄河流域洪水预报现状、存在问题与展望颜亦琪12，王春青 1, 陶新1， 许珂艳1，范国庆1（1 黄委水文局 2 河海大学）摘要：洪水预报是黄河防洪的重要非工程措施，洪水拦、排、分滞的时机和水量都要求精确的洪水预报结果。文章从洪水预报任务、预报模型方法、预报系统建设、预报工作体制等几个方面介绍了黄河流域洪水预报现状及存在问题。也展望了黄河流域洪水预报的发展前景和今后的工作重点。 关键字：黄河流域；洪水预报；现状；存在问题；展望黄河洪水预报现状概述黄河是中国第二大河，是一条著名的多泥沙河流。流域情况十分复杂，其显著特点是：水少沙多，水沙异源，河道迂回宛转，情况复杂多变；春有桃汛，夏秋有伏秋大汛，冬有凌汛；黄河下游河床不断淤积抬高，形成举世著名的地上“悬河”，对下游防汛造成巨大的威胁。黄河流域图见图1.1。洪水预报工作是防洪指挥、调度、决策的重要依据。几十年来，针对黄河的特殊性，黄河几代水文工作者共同研制和引进了多种适用于黄河流域的预报模型方法和手段，并不断进行修正和改进。在历年黄河防汛、水资源统一调度管理中发挥了不可替代的作用。随着黄河的治理开发，对洪水预报提出了更新更高的要求。同时为维持黄河健康生命而实施的一系列治黄实践，如调水调沙、小北干流放淤试验、利用桃汛洪水冲刷潼关高程试验，使黄河洪水预报工作变得更为艰巨，任务多、要求高。图1.1 黄河流域图黄河洪水预报任务区域河名预报站点预报项目黄河上游黄河唐乃亥洪峰流量、峰现时间洮河红 旗湟水民 和大通河享 堂大夏河折 桥黄河中游黄河吴 堡洪峰流量、峰现时间无定河白家川洪峰流量、峰现时间延水河甘谷驿黄河龙 门洪峰流量、峰现时间渭河华 县黄河潼 关黄河三门峡潼关站流量8000m3/s时，制作三门峡水库水位和出库流量预报。黄河小浪底洪峰流量、峰现时间洛河黑石关沁河武 陟黄河下游黄河花园口及其以下六站洪峰流量、峰现时间预报，洪峰水位黄河洪水预报涉及整个黄河流域，包括黄河干支流主要控制站、大型水库、分滞洪区等，预报的重点区域是龙三区间（龙门至三门峡）、三花区间（三门峡至花园口）及黄河下游。河段（区域）预报站点见图1.1，表1.1。预报项目主要有洪峰流量、峰现时间、流量过程、下游各站洪峰水位。近年来，为配合黄河调水调沙、黄河小北干流放淤和利用桃汛洪水冲刷潼关高程试验研究，增加了潼关站、小花区间（小浪底至花园口）日平均流量预报项目以及龙门站洪峰流量、峰现时间、500m3/s流量出现时间和持续时间、次洪水量、最大含沙量等要素的预报工作。预报区域站点及项目见表1.1。表1.1 黄河流域洪水预报任务洪水预报模型方法黄河上、中、下游所使用的洪水预报方法具有一定的区域性差别，这与区域的水文气象特征、资料条件和区域防汛重要程度等有关。目前，黄河洪水预报的主要任务是中下游干支流主要站的洪水预报。黄河上游主要预报方法黄河上游地区为半湿润半干旱地区。暴雨强度较小, 洪水洪峰流量不大, 历时较长, 一般只能形成中下游洪水的基流。黄河上游主要利用降雨产流经验相关，新安江三水源模型，马斯京根流量演算、经验修正等方法进行洪水预报。黄河中游龙三区间主要预报方法黄河中游地区暴雨频繁、强度大、历时短, 形成的洪水具有洪峰高、历时短、含沙量大、陡涨陡落的特点, 是黄河下游的主要成灾洪水。黄河龙三区间洪水预报难点主要有：干支流汇合、无控区间来水及漫滩流量演进。所用的主要洪水预报方法有：（1）洪峰相关法用上游站洪峰或上游干支流相应洪峰流量合成与下游站洪峰建立相关。考虑各站特点，加入各影响因子做为参数。如吴堡站用洪峰涨率参数建立府谷吴堡站相关图，华县站用峰形系数、区域降水、前期影响雨量做为参数建立临潼华县复变数相关图。（2）降雨径流相关及概化单位线对于未控区间，如各支流把口站以下的峡谷区间产生的暴雨洪水用降雨径流关系和概化单位线建立预报图。（3）流量演算一般洪水预报方法根据预报站点上游区间支流汇入情况选择 “先合后演”、“先演后加”或“边演边加”的马斯京根变参数分段流量演算法进行洪水预报。如潼关站，因龙门至潼关区间有渭河、汾河、北洛河三条支流汇入，其中渭河、北洛河是由潼关上游仅2-3km处汇入黄河。在入黄前渭河、北洛河都是单独汇流，互不干扰。因此，采用“先演后加法”进行流量演算。黄河干流龙门的洪水和汾河的洪水在汾河入黄口处汇合，然后再往下游演进。因此，采用“边演边加法”进行流量演算。漫滩洪水预报方法洪水漫滩后，其演进规律也相应地发生了变化，洪峰削减率增大，传播时间延长，具有水库的调蓄特征。因此，根据河道实测淤积断面资料和水位站资料，用“蓄率中线法”分段建立方案，蓄率中线法的实质为图解式的水库调洪演算。并对高含沙大洪水漫滩后泥沙淤积对洪水过程预报的影响建立了实时修正模型。黄河三花区间主要预报方法三花区间洪水预报主要难点：漫滩流量、滩区蓄洪、高水位、传播时间，滞洪区应用后的洪水演进，信息不足、人类活动影响、高含沙水流等。三花区间流域面积较大，自然地理条件比较复杂，加之降雨时空分布很不均匀，因此该区建立的是综合分散性模型，即将全区分块，每块又划分为若干单元，进行产汇流计算。采用的产流模型有降雨径流相关模型、霍顿下渗模型、包夫顿下渗模型、新安江三水源模型和坦克模型，坡面汇流模型为经验单位线，河道汇流模型为马斯京根法。此外，还有水库调洪演算、特殊问题处理及实时校正等计算模型方法。一些分布式流域水文模型也在建设试用中，如全分布式Topkapi水文模型。霍顿模型也称超渗产流模型，在整个三花区间应用情况好于其它模型，这与区间的下垫面条件及降雨特性有关。包夫顿模型也属超渗产流模型，该模型适应于沁河中上游区域。 新安江模型应用在三门峡至小浪底区间和小浪底、黑石关、武陟至花园口区间。针对三花间的特殊问题，有如下处理方法1） 中小水库群处理三花区间有中小水库400多座，这些水库一般只有溢洪口门，且无闸门控制，因此对洪水的影响主要是拦蓄作用，调蓄作用一般很小，水库拦蓄量采用流域填洼公式计算。 2） 滞洪区处理本区有伊洛河的夹滩地区和沁河的沁北两处滞洪区，对于夹滩地区，当伊河龙门镇和洛河白马寺站流量分别超过3000 m3/s时，即有可能决堤滞洪，演算方法有马斯京根法、经验槽蓄法和水库蓄率中线法；沁北滞洪区当沁河流量超过2500m3/s时则自然滞洪，演算方法为马斯京根法。3） 实时校正模型由于流域特性和降雨分布的复杂性和多变性，用降雨径流或河道汇流作出的流量序列预报，有时误差很大，需进行实时校正。本流域采用的是反馈模拟实时校正模型。该模型是充分利用已获得的实测流量信息，并根据这些已出现的实测流量与原预报流量值的关系，对未来的预报流量值进行反馈模拟。黄河下游主要预报方法黄河下游是举世闻名的“地上悬河”, 历史上曾多次发生决口和改道。今天, 黄河下游洪水灾害的隐患依然存在, 主要表现在大洪水时漫堤决口对黄淮海平原的威胁和中常洪水时下游滩区180多万人民生命财产的安全问题。所以，黄河下游洪水预报任务非常重要和艰巨。黄河下游自花园口至河口，全长七百多公里，设有花园口、夹河滩、高村、孙口、艾山、泺口和利津七个水文站，河道由游荡型变为弯曲型。一般洪水由主槽排泄，较大洪水则出槽漫滩。黄河下游洪水预报难点主要为漫滩洪水演进和变动河床水位预报。现在主要用的预报方法有洪峰流量相关和马斯京根流量演算，以及漫滩洪水经验处理方法。还有针对洪水水位预报的多元回归模型、相应水位法、水力学模型、河流动力学模型、人工神经网络方法等。漫滩洪水处理问题是一个难题，根据黄河下游河道滩区的水流特点，黄河水文工作者总结了以下几种洪水漫滩洪水的处理方法：1) 滩区蓄率中线法本法将每段河段的滩地概化为一个线性水库。用马斯京根法做一般洪水演算，发生漫滩洪水时，漫滩流量以上部分再进行水库型调洪演算并扣除损失量，所得结果即为预报成果。2) 滩区汇流系数法对河段中每块闭合（准闭合、非闭合）的滩地分别进行处理，洪水一边向下演进，一边沿途进滩调蓄。本法也是以马斯京根法为基础进行一般洪水演算。发生漫滩洪水时将洪水在入流断面分成滩、槽部分按不同汇流系数分别演算，最后叠加得到预报结果。3) 逐滩演算法河道洪水在自上而下的演进过程中，实际上是不断经过各滩调蓄的。分别对河段中主要滩地的作用进行处理。河槽和滩地均用马斯京根法处理，但演算参数各不相同。洪水流经每个滩地时，将漫滩水流与大河水流在入流断面处分开，分别进行洪水演算，最后在出流断面叠加。 预报手段和预报系统建设 预报手段在洪水作业预报中，由于黄河暴雨洪水的复杂性和特殊性，需要利用多种预报手段进行作业预报，并对各种预报结果进行综合分析、对比和优选。现在黄河流域洪水预报手段主要有：（1）纸质版的黄河流域实用水文预报方案。该汇编选用的预报方案达168个，这些方案在历年洪水预报中发挥了重要作用，至今仍在广泛使用。（2）多个预报软件系统，这些预报系统利用先进的计算机技术将应用在黄河流域的一个或几个预报模型方法集成起来，与实时水情数据库连接，具备信息查询、数据处理、模型率定和洪水预报功能，并有灵活简单的操作界面，提供各式输出。是目前黄河洪水预报的主要手段。 预报系统建设情况：目前，黄河上已建成在使用的洪水预报系统有：黄河洪水预报系统；龙门、潼关、华县站洪水预报系统；小花间分布式水文模型预报系统；黄河下游水位预报系统。渭河分布式水文模型预报系统；基于人工神经网络的黄河下游洪水智能预报系统。其中黄河洪水预报系统是黄河最主要和最常用的洪水预报手段。它的洪水预报覆盖范围为黄河中下游干支流重点河段，实现了黄河潼关三门峡入库、黄河下游花园口以下河道洪水演进及三花区间降雨径流预报。在建的洪水预报系统有：美国可视化河流预报系统（VisualRFS），欧洲水文气象EFFORTS洪水预报系统。待建的洪水预报系统有：黄河中游小流域水沙作业预报系统。先进技术的使用情况：黄河上近年来引进了许多国外先进水文模型，如瑞典的HBV模型，意大利的全分布式Topkapi水文模型，英国的半分布式Topmodel水文模型等。黄河流域在各个洪水预报系统和专家会商系统的建设和使用中大量应用各种先进的计算机手段，如各类计算机交互技术，网络技术，数据库技术等，开发出一些C/S或B/S架构的多功能交互式洪水预报系统，将预报员从烦琐的手工作业中解脱出来。目前，“3S”技术在黄河流域的应用处于初级阶段。GIS目前应用最多，各分布式水文预报系统都基于GIS平台。但RS遥感和GPS全球定位技术目前应用较少，主要有利用雷达和卫星收集空间降水信息。现在黄河流域只有河源区与三花区间有所应用。洪水预报中水文学与其他学科交叉应用技术是未来的发展方向，如与气象、地理、数学、力学、环境学等学科的交叉。因为我们所要预报的对象是一个流域或是一个河段，它们本身不是孤立的，与各种学科因素息息相关。黄河流域在多学科交叉研究方面一直在做着积极的探索。洪水预报工作体制：（1）实行主班预报员制。主班预报员在自己当班时间内时刻关注水情；负责组织洪水预报作业，在预报会商时主讲，为其他预报员提供尽可能多的洪水相关情况，编写水情简报等。（2）实行会商制。洪水预报制作完成之后，必须经过会商后方可对外发布。（3）实行实行预报、审核、签发制。洪水预报发布时应签署预报、审核、签发人姓名。（4）实行考核评分制。存在问题近年来，黄河洪水预报工作在洪水预报模型方法制作、洪水预报系统建设、先进技术使用等方面取得了一定的成就，同时，黄河洪水预报也存在着许多问题：1、洪水预报信息不足，影响洪水预报精度。降雨、水文报汛站网分布不均，一些地区设站稀少，导致信息量不足，报汛时间频次不够也会导致信息量不足，不能满足洪水预报的需要，影响洪水预报精度。黄河洪水测验难度大，预报作业中，很多预报因子无法实时获取。滩区、分滞洪区进出水量、中小水库蓄水情况很难及时、准确得到。2、气象预报与洪水预报尚未有效结合，还没有实现天气、洪水的连续、滚动预报。只有有效结合气象预报与洪水预报，才能有效延长预报期。 3、1990年以后，黄河流域下垫面和河道状况发生了较大改变，应加强对黄河流域产汇流规律、坡面及河道洪水演进规律的研究。4、纸质版黄河流域实用水文预报方案集结了几代水文工作者的心血，应该在对方案进行改进和修正后使其计算机化，进一步为黄河水情服务。5、黄河上游、河三区间洪水预报方法手段太薄弱。6、高含沙洪水预报问题黄河中下游高含沙洪水，在河道演进过程中会发生一些异常变化，如洪峰沿程增加、水位异常偏高或偏低。目前，对高含沙洪水演进的分析还处于现象的解释阶段，认识尚未取得一致。目前，我们还无法预报高含沙洪水现象的发生，高含沙洪水出现后，洪水预报也存在相当大的困难。7、变动河床的水位预报黄河下游变动河床的水位流量关系极为复杂。洪水过程中，过水断面可发生剧烈的冲刷或淤积，很难获取水位预报所需要的一些预报因子。目前，黄河下游水位预报仅取得一些初步成果，还不具备正式发布水位预报的条件。8、流域内多水库联合调度预报方案有待进一步研究。9、泥沙预报，洪水预报大的范畴里应包含泥沙预报。目前，黄河流域泥沙预报只限于经验方法，应加强新方法新模型的研制。10、众多的洪水预报系统各自分散，不利于系统之间的数据共享与传递，各系统预报结果的对比分析，相互修正，统一输出自动化程度不高，大大影响洪水预报作业速度和效率。应最有效集成各系统。展望未来，随着计算机技术、“3S”技术、多学科交叉融合等技术的发展，将涌现出一些结构更趋合理的洪水预报模型。而且，随着对黄河流域产汇流特性、河道坡面洪水演进规律，高含沙洪水演进规律的进一认识，更多的适用于黄河流域复杂情况的洪水预报模型将被研制出来。 随着黄河域站网规划的更趋合理，全国水情分中心的逐步建设完成和自动雨量站的普及、以及各种信息采集、传输手段的现代化，报汛站点空间分布密度和报汛时间频次、精度将能满足分布式洪水预报模型的输入要求。另外GIS技术和水文技术的有效结合，将有望建成黄河中游分布式洪水预报系统。我们甚至可以拭目以待全流域物理概念明确的全分布式气象水文耦合模型的建成。这是黄河洪水预报今后的主要发展方向。洪水预报系统建设将会越来越人性化，最终会实现“我得即我想“功能。总之，随着水文现化化的大力推进，各种先进技术的日益发展，多学科交叉技术的逐步成熟，水文业务队伍素质的不断提高，黄河洪水预报必然会迈上一个新的台阶。以后的黄河洪水预报的工作重点：1、多源数据同化；2、加深认识黄河流域产汇流规律的新变化；3、变动河床水位预报进一步探索；4、黄河不同河道、坡面洪水演进规律深分析，包括漫滩洪水预报研究；5、气象与洪水预报的有效结合，水文气象预报精度提高研究；6、高含沙洪水预报问题研究；7、各洪水预报系统的有效集成；8、气候变化对黄河流域洪水预报的影响研究。参考文献：1 CHENG Zhan-ting, et al. Practical hydrological forecasting methods in the Yellow River Basin M. Zhengzhou, China, Hydrology bureau of Yellow River Conservancy commission, MWR, 1989.2 BAO Wei-ming. Hydrology forecasting M. Nanjing, China, Water Power Press, 2007.3 ZHANG Hong-yue, et al. Hydrologist manual of the Yellow River M. Zhengzhou, China, Hydrology bureau of Yellow River Conservancy commission, MWR, 1991.4 CHENG Xian-de, et al. Hydrology in the Yellow River BasinM. Zhengzhou, China, Yellow River Water Power Press, 1997.5 SHI Fu-cheng, et al. Storms and floods in the Yellow River M. Zhengzhou, China, Yellow River Water Power Press, 1996.6 ZHANG Jian-yun, LIU Jin-ping. Development and Prospects of Hydrological Forcasting Technique in China J. Hydrology, 2005, Vol.25, No.6: 15-19.7 REN Li-liang. Transformation of Hydrological Modeling Techniques in the Digital Era J. JOURNAL OF HOHAI UNIVERSITY, 2000, Vol.28, No.5: 25-30.8 WANG De-jun. Hydrological StatisticalM. Peking, China. Water Power Press,19939 RUI Xiao-fang, HUANG Guo-Ru. The current situation and future tendency of the distributed hydrology model J. Advances in Science and Technology of Water Resources, 2004, Vol.24, No.2: 55-58.10 Abbott M B, Bathurst J C, Cunge J A, et al. An introduction to the European Hydrologic System. System Hydrologique European, SHE J. J of Hydrol, 1986 ,87: 4577.11 Garrote L,ras R L. A distributed model for real-time flood forecasting using digital elevation models J. J of Hydrol, 1995, 167 :279306.12 Ciarpica L,Todini E. TOPKAPI : a model for the representation of the rainfall-runoff process at different scales J .Hydrological Processes, 2002, 16(2): 207229.The Current Situation，Existing Problems and Prospects of Flood Forecasting in Yellow River Basin YAN Yi-Qi12,Wang Chunqing1, Tao Xin1, XU Ke-Yan1, FAN Guo-Qing1(1. Hydrology Bureau of Yellow River Conservancy Commission, Zhengzhou, Henan 450004, China; 2. Hohai University, Nanjing, Jiangsu 210098, China)Abstract: The Yellow River flood forecasting is an important non-engineering measure to control flood. The time and water volume of retaining flood, discharging water and retarding runoff at the detention basins and other work all require the accurate flood forecasting results. This paper introduces the current situation and existing problems of flood forecasting in the Yellow River from the task of flood forecasting, prediction models, forecast system construction, working mode, and other aspects. And also in this paper, the future trend and development in Yellow River flood forecasting was suggested. Keywords: Yellow River Basin; flood forecasting; current situation; existing problems; outlook on future The current situation of Yellow River flood forecasting Overview Yellow River is the second largest river in China and it is a well-known sediment-laden river. Yellow River Basin is very complex, and its distinguishing features are: less water and more sand; heterologous water and sediment; winding riverway, complicated and changeable river regime; there are peach flood in spring, major flood in summer and autumn, icejam flood in winter; the riverbed of downstream Yellow River rises up continuously and eventually become the world-famous hanging river. It throws the enormous threat to flood control.Flood forecasting is an important basis for flood control command, scheduling and decision-making. For decades, in connection with the specificity of the Yellow River, several generations of Yellow River Hydrology workers jointly developed and introduced a wide range of forecasting models and means which are applicable in the Yellow River Basin, and amend and improve these models and means with the passing days. Over the past years, they have played an irreplaceable role in Yellow River Flood Control and water resources management and so on. The development of the Yellow River harnessing put forward higher and newer requirements to the flood forecasting. And to maintain a healthy life of Yellow River, a series of practice are implemented, such as water and sediment regulation, the desilting testing of Xiaobeiganliu, using peach floods scoured Tongguan elevation testing. So there are more tasks and higher requirements to the Yellow River flood forecasting. Figure 1.1 Map of the Yellow River BasinYellow River flood forecasting task Yellow River flood forecasting involves the entire Yellow River Basin, including the control stations of the main reach and some tributaries of the Yellow River, large reservoirs, retarding and detention region. The Long-San region (Longmeng to Sanmenxia), San-Hua region (Sanmenxia to Huayuankou) and the lower reaches of the Yellow River are the key forecasting zone. The Forecast stations are displayed on Figure 1.1 and listed on Table 1.1. The main forecasting items: the peak discharge, the flood peak time, flow process, flood peak water level of the lower reach stations. In recent years, to coordinate with the Yellow River water and sediment regulation, the desilting testing of Xiaobeiganliu and the testing of using peach floods scoured Tongguan elevation, some new items are put forward. Such as the daily average runoff of Tongguan station, Xiao-Hua region (Xiaolangdi to Huayuankou), as well as the flood peak, the flood peak time, the happen time and the duration of 500m3 / s runoff, the largest sediment concentration, the water volume of whole flood process of Longmeng station and so on. The forecast items are listed on Table 1.1.RegionRiver nameForecasting stationsForecasting itemsUpper reachHangheTangnaihaiThe flood peak, the peak timeTaoheHongqiHuangshuiMingheDatongheXiangheDaxiaheZheqiaoMiddle reachHuangheWubaoThe flood peak, the peak timeWudingheBaijiachuanThe flood peak, the peak timeYanshuiGanguyiHuangheLongmengThe flood peak, the peak timeWeiheHuaxianHuangheTongguangHuangheSanmengxiaWhen discharge is above 8000m3/s in Tongguang station, the Sanmengxia reservoir water level and outflow forecasting are needed.HuangheXiaolangdiThe flood peak, the peak timeLuoheHeishiguangQingheWuzhiLower reachHuangheHuayuankou and the other six stations of the lower reachThe flood peak, the peak time , the peak water levelTable 1.1 The main Flood Forecasting task in the Yellow River Basin1.2.Flood Forecasting Models Flood forecasting methods used in the upper reach, middle reach and lower reach have regional differences due to the characteristics of hydro-meteorological, data condition and the importance of the regional flood control and so on. At present, in the Yellow River, the main flood forecast task involves stations on the tributaries and the main stream of the middle and lower reaches.1.3.1 Main forecasting methods on the upper reaches of the Yellow River The upper reaches of the Yellow River belongs to the semi-humid and semi-arid areas with small storm intensity. The flood peak usually lasts a short time, only forms the base flow of the middle and lower reaches. The rainfall-runoff empirical correlating, three-source Xinanjiang model, Muskingum flow routing, the empirical amendment, and other methods are used in the upper reaches flood forecasting. 1.3.2 Main forecasting methods in Long-San region of the Yellow River middle reaches Frequent strong rainfall in the middle reaches always last a short time, which resulting in the flash flood with large peak, short duration and high sediment concentration, the major cataclysm resource of the lower reaches. The distinct flood forecasting difficulties in Long-San region include the convergence of flow from the main stream and tributaries, water from no-control zone and the routing of floodplain flow. The main flood forecasting methods are listed as follows: (1) The flood peak empirical correlating method To build the empirical correlation between flood peak of Upper reach stations （or compositive flood peak from upper main streams and tributaries） and flood peak of the downstream stations. Select some Impact factors as parameters by taking the characteristics of the stations and the river cause into account. For example, Wubao station, select peak rates as a parameter to build the FuguWubao empirical correlating map; while for Huaxian station, select peak shape factor, regional rainfall and antecedent rainfall as parameters to build LintongHuaxian complex variables empirical correlating map. (2) The rainfall-runoff empirical correlating method and generalized unit hydrograph For heavy rain and succedent flood happened in the no-controlled zone, such as the zone from the control stations of the tributaries to the entrance of the main stream. Use the rainfall-runoff empirical correlating method and generalized unit hydrograph to build Forecasting map. (3) Flood routing general flood forecasting methods Select the parameter-varying Muskingum segmentation flood routing with certain mode for flood forecasting according to regime of upper tributaries. The mode includes “composing first, routing second”, “routing first, composing second” and “composing while routing”.For example, above Tongguan station, there are three tributaries: Weihe River, Fenhe River, and Beiluohe River. Weihe River and Beiluohe River separately flow into the Yellow River at the site only 2-3km far away from Tongguang station, without any interference. As a result, select routing first, composing second for flow routing. The mainstream flood from Longmen station and tributary flood from Fenhe River convergent at the entrance of Yellow River. Therefore, select composing while routing mode for flood routing. Over-bank flood forecasting methods After the flood run onto the bank, the routing of the flood will change a lot with reduction of peak rate, prolonging of the spread time. Obviously, the flood plain has reservoir regulation features. For this reason, according to the measured silted across-section data and water level stations data, forecasting model can be build segmentedly by “storage rate ML”. The essence of the “storage rate ML” is schematic reservoir regulating method. 1.3.3 Main forecasting methods in San-Hua region of the Yellow River middle reachesIn this region, the main difficulties in flood forecasting: floodplain flow, flood storage on floodplain, high water levels, spread time, flood routing after the application of the flood detention area, no enough information, the impact of human activities, runoff with lots of sediment. San-Hua region area is large with complicated geographical conditions and uneven distribution of rainfall no matter on space or time. So the integrated disperse model is used in this region. Namely, divide the whole region into several blocks and further divide each block into several units on which flow generating and runoff accumulating are calculated.The flow generation models include rainfall-runoff empirical correlating model, Horton infiltration model, Xinanjiang model and tank model. Slope Runoff concentration model uses empirical unit hydrograph, while river-course runoff concentration model is Muskingum model. In San-Hua region, there are also reservoir flood regulating model, model for solving special problems and real-time correcting model and so on. Some distributed hydrological models are also in the testing and use, such as the Topkapi distributed hydrological model. Horton infil