Algorithms of automatic calibration of multi-parameter models used in operational systems of flash flood forecasting

In the second article of the serie, new tools of fast and effective automatic calibration of multi-parameter hydrological models are presented. In particular, it is shown that the basic method, the Stepwise Line Search (SLS), two its modifications (SLS-F and SLS-2L) and their combination (SLS-2LF) can be used in automated systems of operational flash flood forecasting where computational efficiency is critical due to a great number of catchments that should be calibrated. Results of implementation of the SLS-based calibration procedures for parametrization of the Sacramento Soil Moisture Accounting Model, which demonstrate their high effectiveness and efficiency, are shown.     

集合数值预报方法在山洪预报中的初步应用

选取湖南4个典型山洪个例,分析了嵌套模式不同分辨率的预报效果,并采用多物理过程组合的集合预报方法,研究区域集合数值预报技术在山洪预报中的应用。研究指出,单一模式条件下,使用模式嵌套技术提高分辨率,并不能有效地改进对引发山洪灾害的突发强降水过程的预报能力,且这种单一的确定性预报,对暴雨等灾害性天气的预报存在不确定性。集合预报是解决单一预报不确定性提出的新方法,与单一模式提供的确定性预报结果相比,集合预报可为山洪预报提供多种定量预报产品。利用集合平均预报、极值预报,可以引导对山洪采取分类应对措施;集合预报在给出降水分布的同时,还给出伴随预报结果的可信度;概率定量降水预报(PQPF)与水文预报模型结合,可以用于概率水文预报。     

Evaluating the impact of lightning data assimilation on mesoscale model simulations of a flash flood inducing storm

Flash floods are associated with highly localized convective storms producing heavy rainfall. Quantitative precipitation forecasting of such storms will potentially benefit from explicit representations of deep moist convection in numerical weather prediction models. However, explicit representation of moist convection is still not viable in operational mesoscale models, which rely on convective parameterizations for issuing short to medium-range forecasts. In this study we evaluate a technique that uses regional Cloud-to-Ground (CG) lightning observations to define areas of deep moist convection in thunderstorms and adjust the model-generated precipitation fields in those regions. The study focuses on a major flash flood inducing storm in central Europe (23 August 2005) that was simulated with the aid of an operational weather forecasting system (POSEIDON system based on Eta/NCEP model). The performance of the technique is assessed using as reference distributed rainfall estimates from a network of radar observations. The results indicate that CG lightning data can offer sufficient information to increase the mesoscale model skill in reproducing local convective precipitation that leads to flash floods. The model error correction is shown to be proportional to the density of lightning occurrence, making the technique potentially suitable for operational forecasting of flash flood inducing thunderstorms.     

Applying a coupled hydrometeorological simulation system to flash flood forecasting over the Korean Peninsula

In flash flood forecasting, it is necessary to consider not only traditional meteorological variables such as precipitation, evapotranspiration, and soil moisture, but also hydrological components such as streamflow. To address this challenge, the application of high resolution coupled atmospheric-hydrological models is emerging as a promising alternative. This study demonstrates the feasibility of linking a coupled atmospheric-hydrological model (WRF/WRFHydro) with 150-m horizontal grid spacing for flash flood forecasting in Korea. The study area is the Namgang Dam basin in Southern Korea, a mountainous area located downstream of Jiri Mountain (1915 m in height). Under flash flood conditions, the simulated precipitation over the entire basin is comparable to the domain-averaged precipitation, but discharge data from WRF-Hydro shows some differences in the total available water and the temporal distribution of streamflow (given by the timing of the streamflow peak following precipitation), compared to observations. On the basis of sensitivity tests, the parameters controlling the infiltration of excess precipitation and channel roughness depending on stream order are refined and their influence on temporal distribution of streamflow is addressed with intent to apply WRF-Hydro to flash flood forecasting in the Namgang Dam basin. The simulation results from the WRF-Hydro model with optimized parameters demonstrate the potential utility of a coupled atmospheric-hydrological model for forecasting heavy rain-induced flash flooding over the Korean Peninsula.     

Development and Application of an Atmospheric-Hydrologic-Hydraulic Flood Forecasting Model Driven by TIGGE Ensemble Forecasts

A coupled atmospheric-hydrologic-hydraulic ensemble flood forecasting model,driven by The Observing System Research and Predictability Experiment (THORPEX) Interactive Grand Global Ensemble (TIGGE) dat...     

山洪灾损预估方法研究

为了预估四川省的山洪灾害造成的人口和经济影响,本文基于四川省数字高程模型数据(DEM),依据山洪沟的判别标准,提取出全省范围的山洪沟,并结合坡度与降水耦合的关系建立了四川山洪危险性评价模型。以2011年1月~2013年12月发生的21次山洪灾害为样本,利用2013年四川省人口和经济数据,建立了山洪灾害对人口和经济影响的预估模型。结论如下:(1)提取出影响生命财产安全的山洪沟64346条,其主要分布在盆周山区、川西高原和攀西地区,最大影响面积约38万km2,约占全省面积的78%;(2)建立了山洪危险性评估模型,评估结果能客观地预警四川范围内的山洪等级和影响范围;(3)建立了山洪灾害影响预估模型,模型对人口的影响预估结果比较可靠。      

集合数值预报在洪水预报中的应用进展

水文集合预报是近几年正在形成和发展的水文预报分支,其发展大致可分为两个阶段:第1阶段是1970年至20世纪末进行的长期径流预报,第2阶段从21世纪开始,主要学习气象数值预报中集合预报的概念在短期水文集合预报中的应用。目前,除了单一预报中心的集合预报系统在水文集合预报中应用外,多个预报中心的集合预报大集合也逐渐被应用于流域水文预报,甚至一些小流域的洪水预报。如利用TIGGE(THORPEX Interactive Grand Global Ensemble)集合预报驱动形成的大气-水文-水力的串联系统进行早期的洪水预警研究,将全球集合预报作为洪水模型输入的有限区域模式的初始条件和侧边界条件的研究。这些均表明,基于水文集合预报的洪水预报增加了预报附加值,并能够延长预警提前时间。以欧洲中期天气预报中心的欧洲洪水预警系统(EFAS)和美国NOAA的先进水文预报系统(AHPS)为代表,实现了集合预报在洪水中的实时业务预报,但仍存在数据处理和计算量大,以及如何基于集合水文预报做决策等问题。对于水文集合预报的前处理和后处理的各种技术已处于探索和验证阶段,如何更好地理解基于概率预报的洪水预警决策仍存在许多困难和挑战。     

基于隐马尔可夫模型的中小河流致灾雨量阈值研究

基于数字高程模型(digital elevation model,DEM)高程数据和D8算法实现了湖南中小河流流域子流域划分。利用洪水预报网络传播模型,构建中小河流各子流域径流拓扑结构并提出了空间子流域等效面雨量的概念。基于等效面雨量序列和径流观测序列,构建了基于时空概念的隐马尔可夫降水-径流模型,并利用该模型计算了中小河流各子流域不同时间尺度条件下暴雨致灾临界面雨量阈值,最后利用2010—2015年实况资料对阈值进行检验,结果证明相比传统统计法,新方法的计算结果与传统方法结果一致且具有很好的准确性和稳定性。基于2015年6月的一次暴雨灾害预报,证明了该方法适应于业务化运行。     

Post-processing and forecasts updating in automated systems of flash flood forecasting

This paper completes the series of three articles devoted to automated forecasting of flash floods [3, 5] and describes an effective approach of forecast updating through post-processing operations, which can be useful only in conjunction with such fast and efficient real-time re-calibration algorithms as SLS-based methods are. In particular, a proposed methodology is aimed to reduce negative consequences of scarce or low-quality data that can corrupt optimized parameters and, therefore, lower forecasting efficiency. A new modification of SLS-based optimization that supposes simultaneous re-calibration of the model and correction of the model input by generating of ensemble noises (SLS-E) is presented.     

A case study of excessive rainfall forecasting

Summary Flash floods have been recognized as one of the most significant natural disaster problems in the world. Within the United States, the annual average flood death toll exceeds one hundred and property damage is on the order of a billion dollars. There has been an increased effort of the meteorological community to improve short term quantitative precipitation forecasting, principally by improving mesoscale numerical weather prediction for heavy rain events. Nevertheless, to date, numerical weather prediction has had rather limited impact on the prediction of the most damaging convective rainstorms.This study examines numerical experiments, including both coarse-mesh and fine-mesh model simulations, of the Enid, Oklahoma flood of 10–11 October 1973. Besides the great concentration of rainfall, the Enid flood was rather unique in comparison with other flash flood cases in that it was part of a much larger area of heavy rainfall which soaked the central Plains over the 24h period ending at 1200 UTC 11 October. The objective is to assess the overall usefulness and limitation of numerical weather prediction models in quantitative precipitation forecasting for this flash flood event.The model experiments reveal that the broad-scale precipitation patterns associated with the front and cyclone are well predicted, but the maximum rainfall amounts around Enid are underpredicted. The fine-mesh model is superior to the coarse-mesh model because of the former's ability to generate many significant mesoscale features in the vicinity of the front. In the fine-mesh model, many convection-related parameters (e.g., moisture flux convergence) are correlated very well temporally and spatially with the observed heavy precipitation scenario.     

Storm Tracking and Monitoring Using Objective Synoptic Diagnosis and Cluster Identification from Infrared Meteosat Imagery: A Case Study

Summary  The present paper investigates the potential of combining image processing techniques based on cluster analysis of infrared (IR) Meteosat images with dynamic meteorological theory on synoptic systems. From this last point of view the highest probability of deep convective development is favoured where the overlapping of four mechanisms acting at synoptic scale is produced: upward quasi-geostrophic forcing, convergence of water vapour at low levels, convective instability in the lower troposphere and great convective available potential energy. Cloud tracking is performed over sequences of Meteosat IR images by using a shape parameterisation approach after appropriate filtering for non-significant clouds and automated identification of convective systems. The integrated methodology is applied to the case study of the heavy rainfall event which produced floods in the South of France and the North of Italy on September 27–28^th, 1992. The analysis focuses on the monitoring and explanation of the zones most affected by heavy rainfall with the aim of investigating possible improvements of the predictive potential of cloud tracking and allowing identification of the areas which most lend themselves to flash floods for use in operational flood forecasting applications. Received July 20, 1998/Revised June 21, 1999     

基于非静力模式物理扰动的中尺度集合预报试验

以GRAPES中尺度有限区模式作为试验模式, 从模式的不确定性方面来构造中尺度的集合预报, 重点考虑物理因子与初始条件的扰动作用。针对2004年7月10日北京城区的突发性暴雨过程进行了36 h的集合预报试验。结果表明:GRAPES模式可有效地捕捉到中尺度过程的信息; 中尺度集合预报是可行的, 可改进中尺度暴雨过程落区、强度的预报; 不同集合方案的预报结果各不相同, 同一方案各个成员的预报结果也有差异, 即存在适宜的离散度; 在离散度分析中发现在北京附近存在一个明显大值区, 且在大气中低层的垂直结构表现出一致性, 表明这一区域的预报不确定性很大。从集合检验结果中得到:单纯考虑模式物理扰动来构造中尺度集合预报系统有一定难度, 当加入初始场不确定信息后, 同时考虑模式的不确定性和初始场的不确定性, 有助于捕捉更多的中尺度系统的不确定信息, 有助于构造更为有效的中尺度集合预报系统。     

气象水文模型耦合在黄河三花间洪水预报中的应用

水文预报的有效预见期越长,预报精度越高,其对防洪减灾的价值就越大,而气象水文模型的耦合是延长水文预报有效预见期的一个重要研究方向。选取黄河流域三门峡水库以下、花园口水文站以上集水面积为研究区域,基于空间分辨率为90 m×90 m的数字高程模型数据构建数字流域水系,利用中尺度非静力模式MM5进行降雨预报和气温预报,并将之作为分布式时变增益水文模型降雨和蒸发计算的输入进行关键站点的水文预报,实现了分布式时变增益水文模型与大气中尺度非静力模式MM5的单向耦合及定量降水预报与洪水预报的结合,为花园口防汛工作提供了一定的参考价值。2012年与2010年的流量滚动预报结果相比,2012年预报结果的效率系数较大,径流深相对误差较小,说明2012年预报效果较好。对预报结果分析表明,实测的水情资料完整性直接影响预报结果,而滚动预报时间段的增加,水文预报的初始场对径流预报的影响逐渐减弱,水文预报的精确度逐渐提高,预报的效果变好。     

基于HBV模型的淮河流域洪水致灾临界雨量研究

根据流域暴雨洪水致灾机制,文章提出了考虑前期基础水位的动态致灾临界雨量指标,并以淮河上游地区为例,基于HBV水文模型建立了降水-流量-水位关系,并根据这种关系确立了临界雨量确定的方法流程.首先基于历史水文数据率定和验证模型,得到适用于研究区的最优化模型参数,然后构建洪水上涨期水位流量关系,最后以是否达到致灾水位为标准,通过模型试算并结合水位流量关系曲线反推出致灾临界雨量值.在淮河上游地区的研究中,利用2002-2009年逐日气象水文数据对HBV模型进行了参数率定和检验,并针对洪水过程进行了参数优化,经过率定后HBV模型对王家坝以上流域具备较好的适用性,对典型洪水过程模拟的确定性系数和NASH效率系数均在0.8以上;根据王家坝站实测流量水位数据,构建了概化的单一关系曲线;结合HBV模型和水位流量关系得到了王家坝以上流域的动态致灾临界雨量指标,临界雨量值随前期基础水位升高而减小,并且随着前期水位的变化,临界雨量值呈现了明显的非线性响应特征.     

暴雨致洪预报系统及其评估

该文研发了基于CREST V2.1分布式水文模型的暴雨致洪预报系统,应用中国气象局降水业务产品,开展全国0.125°×0.125°的逐日洪水预报和区域30"×30"的逐时洪水预报。其中,全国洪水预报以松花江、辽河、海河、黄河、淮河、长江、东南诸河、珠江、西南诸河和西北诸河十大水资源分区的典型流域为研究对象,区域洪水预报以淮河流域为研究对象。以模拟和观测流量之间的效率系数为目标函数,采用SCE-UA方法分别对全国和区域洪水预报模型的参数分流域进行率定。评估参数率定前后模型对效率系数、相关系数和相对偏差的改进程度,并对参数率定后的模型进行检验。结果表明:率定后的模型能够重现控制水文站的实测洪水过程,与率定前相比,效率系数和相对偏差有显著改进,相关系数有较大改进。系统符合业务需求,具有较好的预报精度和时效性,具备业务应用能力。     

基于BP和Elman神经网络的福建省汛期旱涝预测模型

建立了福建汛期旱涝BP和Elman神经网络预测模型，并对两种模型的性能和差异进行了比较，结果表明：动量BP网络模型，特别是具有局部反馈特性的Elman网络模型具有较好的拟合精度和预报效果。此外两种模型对旱涝等级为2和4的预测偏差较大，而对旱涝等级为3的预测较为准确。     

北京地区城市暴雨积涝灾害风险预评估

该文提出自下而上的城市暴雨积涝灾害风险定量评估方法,即在三级评估指标体系下,由下级指标综合核算上级指标系数。在第2级指标计算中,风险区划的危险性指数由历史降水量资料推算得出,风险预警则用实况及预报降水量来计算致灾因子危险性指数;暴雨敏感性指数综合叠加地形、不透水地表因子及河网密度得出;暴雨积涝的风险暴露因子侧重地均人口密度、地均GDP及重点防汛指标等因子,着重于城市地区人口、经济、防汛重点目标的暴露程度。然后在危险性、敏感性及暴露性指数的基础上叠加得出积涝风险指数。通过对比发现,得到的风险区划结果与2004—2008年北京地区暴雨积涝的历史灾情基本吻合。最后,选用北京2011年“6.23”暴雨作风险预警的实例应用检验及分析,结果表明:采用自下而上的快速风险评估结果与积涝的实际发生情况较为接近,无论是风险变化趋势还是风险区域分布情况均与当天的积涝发生情况基本吻合。即该方法能较为准确、快捷地圈定城市地区各级风险区域,能较好地满足风险评估、区划及风险预警的要求。     

What spatial resolution do we need for a route-based road weather decision support system?

Since their implementation, road weather information systems have mostly relied on point measurements from outstations to initiate and verify daily forecasts. Initially, spatial extrapolation was achieved by thermal mapping, but this is gradually being replaced by route-based forecasting techniques. Both techniques are similar in the sense that they use a point measurement, often taken from an outstation, to provide a spatial forecast of road surface temperatures around the road network at varying resolutions. A substantial research effort has been undertaken to understand and model the complex environmental conditions and mechanisms responsible for the variation in road surface temperatures around the road network. In particular, the interaction of varying geographical parameters around the road network (e.g. altitude, land use, road construction, topography, etc.) has been used to develop local climatological models and route-based forecasting products. By considering the needs of winter maintenance engineers, this paper reviews the current state of the art and takes a critical look at the embedding of forecast products into decision support systems. This is achieved by considering a case study of how road surface temperature and condition vary across the width of a road profile, instead of just lengthways along a road. It is shown that temperature and condition both vary significantly across the profile, which immediately raises questions about the validity of current surveying and modelling practices. This has implications for both the resolution of route-based forecasting products as well as user confidence in automated decision support systems.     

基于NCEP/GFS资料的中国东部地区雷暴预报研究

基于来自美国国家环境预测中心(NCEP)的GFS(Global Forecasting System)分析及预报场资料,将多个能够表征雷暴发生动力、热力环境的对流因子作为预报因子,通过费希尔判别准则及逐个引入因子法,建立集合多个对流参数的雷暴预报模型,从而进行较长时效(12—24 h)的区域性雷暴预报。依据临界成功指数(CSI)最高的原则,建立最优预报模型,不同地区所选用的对流参数不同,雷暴模型预报雷暴发生与否的临界值也不同,从而不仅能够得到较好的集合多个对流参数的雷暴区域性预报,还能充分考虑不同地区雷暴发生的地域性特点和气候背景。将建立的预报方法应用于2012年6和9月的两次强对流过程的预报,发现雷暴预报模型较好地预报出两次过程的雷暴落区。进一步,为了能够在强天气预报中客观有效地区分出雷暴与暴雨区,引入集合动力因子暴雨预报方法。集合动力因子暴雨预报方法在诊断和追踪强降水的发展演变中表现凸出,而集合对流参数雷暴预报方法则对包含短时强降水、冰雹、大风等在内的对流性天气有较好反映,综合两套预报方法各自的优势,建立了集成动力因子-对流参数强天气预报方法,用于降水和雷暴的预报,同时对雷暴加降水型、雷暴无降水型、降水无雷暴型等强天气进行区分预报。对中国多个典型城市的预报效果分析发现,该方法不仅能够较好地预报出较长时效(24 h)的雷暴和降水落区,对区分降水雷暴、无降水雷暴和无雷暴降水也表现出一定的能力。     

热带气旋路径趋势数值预报的试验研究

给出了一个嵌套于欧洲中期天气预报中心全球谱模式中的套网格模式，用于预报热带气旋路径趋势，提出了一种以欧洲中期天气预报中心提供的时距为24小时的预报场为基础的“接力预报”方法；设计了五种不同的“接力预报”方案；并对8507号和8509号两个热带气旋就各种不同预报方案进行了数值预报试验。结果表明，本模式对热带气旋路径趋势具有较强的预报能力，对疑难路径也具有一定的预报能力，为开展热带气旋的路径趋势预报提供了一个新的途径。文中还对影响热带气旋路径趋势的一些主要因子结合各预报结果进行了分析，为模式尽早投入业务提供了一定的依据。