降水空间不均匀性对径流过程模拟的影响

应用数字高程流域水系模型，通过子流域和网格2种空间离散方式，采用传统的子流域法、雨量插值子流域法、逐个网格法3种雨量数据输入处理方法，分别作日流量模拟与洪水过程模拟，模拟过程与实测水文过程相比较表明：日模中3种雨量处理方法计算精度相当，次模中逐个网格法优于传统子流域法和雨量插值子流域法；降雨空间分布不均匀性考虑得越充分，水文过程模拟精度越高。     

雨量站网布设对水文模型不确定性影响的比较

雨量站网布设会影响径流模拟精度,研究不同雨量站密度和空间分布的径流响应规律对提高径流模拟精度和减小不确定性具有重要意义。应用新安江模型和HBV（Hydrologiska Fyrans Vattenbalans）模型,以湘江流域为研究对象,采用贝叶斯方法比较分析在不同雨量站密度及空间分布下径流模拟的不确定性。结果表明:增加雨量站密度可以降低面雨量的估计误差,使模型在不同的雨量站空间分布下具有较高的模拟精度;通过优化雨量站空间分布,可以减小雨量站网布设导致的模型不确定性,从而提高径流模拟精度;在相同的降雨输入和参数采样方法下,新安江模型和HBV模型对降雨输入导致的不确定性响应规律具有相似性,但是本研究结果显示在湘江流域新安江模型的模拟精度更高,而HBV模型的不确定性更大。     

基于不同雨型的裴河小流域山洪灾害临界雨量分析

临界雨量是山洪灾害预警的重要指标,为分析小流域雨型对临界雨量的影响,首先对流域降雨过程按照雨峰出现位置的不同进行归类和划分,定性分析流域常出现的各种典型雨型;在此基础上,采用PilgrimCordery法定量确定流域各种典型雨型的时程分配,并通过试算法计算其相应的临界雨量;利用流域近期发生的成灾洪水,采用临界雨量偏离度指标进一步分析论证雨型对临界雨量的影响。以河南省新县裴河典型小流域为实例进行应用研究,结果表明:设计雨型不能完全代表该流域的降雨类型,其属于雨峰偏后式雨型,由此确定的临界雨量与该流域常出现的雨峰偏前式雨型临界雨量相差33%,与多峰雨型临界雨量差异较大,相差43%。实时预警时,根据实时雨型类型,采用相应的临界雨量,可以提高山洪灾害预警预报水平。     

基于深度学习的雷达降雨临近预报及洪水预报

为探究深度学习的雷达降雨临近预报在流域洪水预报中的适用性,采用U-Net、嵌入注意力门的Attention-Unet和添加转换器的多级注意力TransAtt-Unet开展雷达降雨临近预报,将预报降雨作为HEC-HMS水文模型的输入,对柳林实验流域进行洪水预报。结果表明：1 h预见期时,Attention-Unet对短时强降雨预报结果较好,TransAtt-Unet预报降雨模拟的洪峰流量和径流量相对误差小于20%,各深度学习模型对量级较大的降雨和洪水预报精度较高;2 h预见期的预报降雨强度、降雨总量、洪峰流量和径流量存在显著低估,U-Net能取得相对较好的降雨预报结果。基于深度学习的1 h预见期雷达降雨临近预报及洪水预报可为流域防洪减灾提供科学依据。     

降雨输入不确定性对分布式流域水文模拟的影响研究——以武烈河流域为例

降雨输入对分布式流域水文模拟具有重要影响。针对流域降雨资料不完整的情况,以武烈河流域为例,基于反距离加权平均法对雨量站降雨资料进行插补延长,并结合SWAT模型研究了降雨输入不确定性对分布式流域水文模拟的影响。结果表明:不同降雨输入对流域平均降雨量的影响较小,但基于气象站资料的降雨数据在降雨空间差异显著的年份会明显低估面雨量,且在夏季汛期表现更为显著;不同降雨输入对分布式流域水文模拟的影响较大;在雨量站降雨资料不完整的情况下,通过对雨量站降雨数据进行插补延长,相对于直接利用气象站降雨资料,在一定程度上可以提高径流模拟精度,满足降雨资料欠缺流域分布式水文模拟的实际需求。     

西江流域灾害性暴雨洪水特征及成因分析

为探究西江流域暴雨洪水发生规律,根据1994、1998、2005和2008年四场灾害性暴雨洪水实测资料,对暴雨洪水过程进行分析。通过计算暴雨时间、空间变差系数、相对中心和洪水集中度等特征指标定量分析暴雨洪水特性,结合洪水组合和遭遇情况分析洪水成因。分析结果表明:四场暴雨雨量分布不均,降雨历时均大于10d,且暴雨相对中心值呈减小趋势,暴雨中心沿河流流向移动易导致灾害性洪水。1998和2005年两场100年一遇洪水均为全流域大量级洪水遭遇导致。两场暴雨均有雨量大、时间变差系数V_t值小,时间分布均匀和雨峰系数C_p值大,主雨峰峰现时间迟的特点;暴雨中心均有向下游转移的趋势,空间变差系数V_p值均趋于增加,空间分布趋于集中。研究结果可为西江流域的水库群联合防洪调度提供科学依据。     

大尺度流域基于站点的降雨时空展布

降水是影响流域水循环最活跃的因素,其时空分布不均匀对流域产汇流的影响非常大,因此在进行流域水文模拟时,需要充分考虑到降雨的时空变异性。本文针对大尺度流域的特点,提出了一种日雨量的空间展布方法,该方法根据站点降雨量之间的相关系数选取参证站点,采用距离平方反比法把站点的日雨量插补到空间计算单元上去,或采用泰森多边形法插值。该方法计算结果和本次全国水资源规划的结果差别不大,说明插值方法比较可靠。由于大尺度流域雨量站点的布设比较稀疏,采用空间插值的方法生成计算单元降雨量值精度不好,本研究中提出了一个大于10mm的大强度日降雨的向下尺度化方法。首先建立了一个将日降雨分配成小时降雨量的雨强历时关系模型：然后根据黄河流域的特点,把全流域划分为五个大强度降雨区,分区对参数进行了率定：最后采用实测资料对模型进行了检验。本文提出的方法,在黄河流域水文模拟中进行了应用,由径流模拟结果看,效果令人满意。     

基于地理空间要素的雅砻江流域面雨量估算

反映流域整体降水情势的面雨量一直是水文模型的重要输入参数之一,在泰森多边形雨量法的基础上考虑地理空间要素对降雨空间分布的影响,采用面向对象的遥感信息聚类方法提取出雅砻江流域2项形状因子(周长、面积)和5项地形因子(平均高程、平均坡度、平均坡向、高程差周长比和高程差面积比)。降雨—径流相关性分析结果表明:地形因子雨量法在月尺度上的降雨估算精度高于年尺度,且在月尺度上能更好地反映流域不同区段的降雨空间分布特征;在月、年际降雨变化趋势分析方面,年尺度上的降雨与径流一阶差分后平均相关系数为0. 903,高于月尺度的0. 629,主要由于水电站调蓄过程对流域径流异质性的影响,且影响度随着时间尺度缩小而放大。     

岳城水库控制流域暴雨洪水的时程分布规律及分期划分研究

从两个方面全面分析了岳城水库控制流域的暴雨洪水时程分布及分期规律:一是根据实测水文资料,通过统计暴雨日数,典型历时年最大降水量、年最大洪峰流量和5日洪量在各旬出现的频率,对流域暴雨洪水时程分布特性进行了多方面的综合分析;二是通过对流域所处地区暴雨形成的天气成因分析,为前述的统计分析结果提供必要的成因基础.还利用旬平均降雨量的相对系数来确定汛期的起止时间.分析结果表明,岳城水库控制流域的暴雨洪水时程分布具有比较明显的变化规律,其中有一个时期暴雨洪水出现的次数多、量级大,主汛期特征显著,汛期可以划分为3个分期,即前汛期、主汛期和后汛期.     

开都河流域山区径流模拟及降雨输入的不确定性分析

选取塔里木河源区的开都河流域为研究区,将流域内气象水文站点数据与遥感数据相结合,利用气象、土壤类型、土地利用和地表覆盖、数字高程(DEM)和降雨等资料,模拟流域水文过程,并在出山口实测径流数据的基础上对模型进行率定和验证;对降雨输入所带来的径流模拟不确定性进行分析,探讨降雨输入的空间异质性对水文预报结果的影响机制.结果表明:MIKE-SHE模型能在水文、气象站点稀少,土壤及水文地质数据缺乏的条件下,模拟开都河流域的日径流过程,但精度上仍有待提高;降雨输入的空间分布程度对径流模拟有重要影响.FY-2C遥感估算降雨资料能够更好地表达降雨时间的空间异质性,相应地对径流模拟精度也有一定程度的提高.     

Probabilistic disaggregation of a spatial portfolio of exposure for natural hazard risk assessment

In natural hazard risk assessment situations are encountered where information on the portfolio of exposure is only available in a spatially aggregated form, hindering a precise risk assessment. Recourse might be found in the spatial disaggregation of the portfolio of exposure to the resolution of the hazard model. Given the uncertainty inherent to any disaggregation, it is argued that the disaggregation should be performed probabilistically. In this paper, a methodology for probabilistic disaggregation of spatially aggregated values is presented. The methodology is exemplified with the disaggregation of a portfolio of buildings in two communes in Switzerland and the results are compared to sample observations. The relevance of probabilistic disaggregation uncertainty in natural hazard risk assessment is illustrated with the example of a simple flood risk assessment.     

Flood forecasts based on multi-model ensemble precipitation forecasting using a coupled atmospheric-hydrological modeling system

The recent improvement of numerical weather prediction (NWP) models has a strong potential for extending the lead time of precipitation and subsequent flooding. However, uncertainties inherent in precipitation outputs from NWP models are propagated into hydrological forecasts and can also be magnified by the scaling process, contributing considerable uncertainties to flood forecasts. In order to address uncertainties in flood forecasting based on single-model precipitation forecasting, a coupled atmospheric-hydrological modeling system based on multi-model ensemble precipitation forecasting is implemented in a configuration for two episodes of intense precipitation affecting the Wangjiaba sub-region in Huaihe River Basin, China. The present study aimed at comparing high-resolution limited-area meteorological model Canadian regional mesoscale compressible community model (MC2) with the multiple linear regression integrated forecast (MLRF), covering short and medium range. The former is a single-model approach; while the latter one is based on NWP models [(MC2, global environmental multiscale model (GEM), T213L31 global spectral model (T213)] integrating by a multiple linear regression method. Both MC2 and MLRF are coupled with Chinese National Flood Forecasting System (NFFS), MC2-NFFS and MLRF-NFFS, to simulate the discharge of the Wangjiaba sub-basin. The evaluation of the flood forecasts is performed both from a meteorological perspective and in terms of discharge prediction. The encouraging results obtained in this study demonstrate that the coupled system based on multi-model ensemble precipitation forecasting has a promising potential of increasing discharge accuracy and modeling stability in terms of precipitation amount and timing, along with reducing uncertainties in flood forecasts and models. Moreover, the precipitation distribution of MC2 is more problematic in finer temporal and spatial scales, even for the high resolution simulation, which requests further research on storm-scale data assimilation, sub-grid-scale parameterization of clouds and other small-scale atmospheric dynamics.     

新疆阿勒泰地区的洪水特性

受洪水补给来源和阿尔泰山山势影响,阿勒泰地区洪水发生时间比新疆其他以高山永久积雪和冰川融水补给为主的河流偏早。因此,从认识洪水特性、预防洪水灾害角度出发,对阿勒泰地区洪水发生机制、时空分布和变化特点进行分析研究是很有必要的。利用阿勒泰地区各主要河流1960~2011年历年洪峰、冬季和汛期降水、气温和700hPa高空温度资料,对洪水发生机制、时空分布和变化特点进行分析。结果表明:(1)本区洪水类型有融雪型洪水、融雪与降水混合型洪水和暴雨型洪水;(2)洪水与冬季积雪和汛期气温、降雨有关;(3)汛期过程相对短,主汛期发生时间以5月下旬至6月中旬期为主;(4)洪水挟沙能力强,汛期含沙量占年含沙量近90%;(5)暴雨洪水多发生于阿勒泰地区的中东部区域河流上;(6)全区洪水具有同步性,洪水年际变化相对小。     

基于MIKESHE分布式水文模型的降水时间尺度对喀斯特流域径流模拟的影响研究——以红水河系六硐河流域为例

径流形成是流域系统对降水水文响应的结果,其实质也是流域对降水时间和空间上的重新分配。降水作为径流模拟中最大的不确定因素之一,其时空分布特征是影响径流模拟及一系列其它水文问题的主要控制因素。为了探索喀斯特流域降水时间尺度对径流的影响规律,本研究利用MIKESHE模拟技术,结合喀斯特流域独特的地质、地貌、覆被类型及水文特征,对模型的相关参数进行率定,模拟了5场典型洪水的降水径流过程。并且通过改变降水资料的时间尺度,研究降水时间变异对模拟洪水过程的影响。研究表明,在相同的流域下垫面和模型条件下,若累计降水时间增加,同样数量的降水产生的径流总量也会增加,最大的流量出现时间也相对滞后。      

Probabilistic Fuzzy Logic Modeling: Quantifying Uncertainty of Mineral Prospectivity Models Using Monte Carlo Simulations

Significant uncertainties are associated with the definition of both the exploration targeting criteria and computational algorithms used to generate mineral prospectivity maps. In prospectivity modeling, the input and computational uncertainties are generally made implicit, by making a series of best-guess or best-fit decisions, on the basis of incomplete and imprecise information. The individual uncertainties are then compounded and propagated into the final prospectivity map as an implicit combined uncertainty which is impossible to directly analyze and use for decision making. This paper proposes a new approach to explicitly define uncertainties of individual targeting criteria and propagate them through a computational algorithm to evaluate the combined uncertainty of a prospectivity map. Applied to fuzzy logic prospectivity models, this approach involves replacing point estimates of fuzzy membership values by statistical distributions deemed representative of likely variability of the corresponding fuzzy membership values. Uncertainty is then propagated through a fuzzy logic inference system by applying Monte Carlo simulations. A final prospectivity map is represented by a grid of statistical distributions of fuzzy prospectivity. Such modeling of uncertainty in prospectivity analyses allows better definition of exploration target quality, as understanding of uncertainty is consistently captured, propagated and visualized in a transparent manner. The explicit uncertainty information of prospectivity maps can support further risk analysis and decision making. The proposed probabilistic fuzzy logic approach can be used in any area of geosciences to model uncertainty of complex fuzzy systems.     

Characteristics of agro-meteorological disasters and their risk in Gansu Province against the background of climate change

A formal Bayesian approach that uses the Markov chain Monte Carlo method to estimate the uncertainties of natural hazards has attracted significant attention in recent years, and a fuzzy graph can be considered an estimation of the relationship that we want to know in risk systems. However, the challenge with such approaches is to sufficiently consider uncertainty without much prior knowledge and adequate measurement. This paper proposes a new adaptive Bayesian framework that is based on the conventional Bayesian scheme and the optimal information diffusion model to more precisely calculate the conditional probabilities in the fuzzy graph for recognizing relationships and estimating uncertainty in natural disasters with scant data. This methodology is applied to study the relationship between the earthquake’s magnitude and the isoseismal area with strong-motion earthquake observations. It is also compared with other techniques, including classic Bayesian regression and artificial neural networks. The results show that the new method achieves better performance than do the main existing methods with incomplete data.     

黄河源区雨季降水与汛期径流量重建及其千年尺度下的演变特征

黄河源区年内降水集中, 洪水风险大, 重建源区雨季降水和汛期径流量对于提高径流预报预测精度及防洪防灾具有重要科学意义和应用价值。本文利用黄河源区及周边筛选的16个树轮年表, 采用嵌套主成分层次贝叶斯回归模型, 估算参数后验分布替代固定值以考虑不确定性, 重建了黄河源区过去1 160 a的雨季降水; 提出了基于年径流的分类占比回归模型, 以考虑汛期径流量与年径流量的一致性, 将黄河源区汛期径流量展延至公元159年。结果表明: ①嵌套主成分层次贝叶斯回归模型的误差缩减值(E_R)和有效系数(E_C)评价指标值均显著高于0, 分类占比回归模型的E_R和E_C值最高分别达0.90和0.88, 重建结果可靠性较高; ②即使在千年尺度下, 1979—1985年亦是较为不寻常的汛期高径流量时期。     

Impact of Climate Change on the Regional Hydrology – Scenario-Based Modelling Studies in the German Rhine Catchment

The aim of the study is an impact analysis of global climate change on regional hydrology with special emphasis on discharge conditions and floods. The investigations are focussed on the major part of the German Rhine catchment with a drainage area of approx. 110,000 km². This area is subdivided into 23 subcatchments. In a first step, the hydrological model HBV-D serves to simulate runoff conditions under present climate for the individual subbasins. Simulated, large scale atmospheric fields, provided by two different Global Circulation Models (GCMs) and driven by the emission scenario IS95a (“business as usual”) are then used as input to the method of expanded downscaling (EDS). EDS delivers local time series of scenario climate as input to HBV-D. In a final step, the investigations are focussed on the assessment of possible future runoff conditions under the impact of climate change. The study indicates a potential increase in precipitation, mean runoff and flood discharge for small return intervals. However, the uncertainty range that originates from the application of the whole model chain and two different GCMs is high. This leads to high cumulative uncertainties, which do not allow conclusions to be drawn on the development of future extreme floods.     

The use of the possibility theory to investigate the epistemic uncertainties within scenario-based earthquake risk assessments

This paper presents a methodology to represent and propagate epistemic uncertainties within a scenario-based earthquake risk model. Unlike randomness, epistemic uncertainty stems from incomplete, vague or imprecise information. This source of uncertainties still requires the development of adequate tools in seismic risk analysis. We propose to use the possibility theory to represent three types of epistemic uncertainties, namely imprecision, model uncertainty and vagueness due to qualitative information. For illustration, an earthquake risk assessment for the city of Lourdes (Southern France) using this approach is presented. Once adequately represented, uncertainties are propagated and they result in a family of probabilistic damage curves. The latter is synthesized, using the concept of fuzzy random variables, by means of indicators bounding the true probability to exceed a given damage grade. The gap between the pair of probabilistic indicators reflects the imprecise character of uncertainty related to the model, thus picturing the extent of what is ignored and can be used in risk management.     

Managing expert-information uncertainties for assessing collapse susceptibility of abandoned underground structures

Assessing the collapse susceptibility of abandoned cavities at a regional scale is associated with large uncertainties that are mainly related to the very nature of the phenomena, but also to the difficulty in collecting exhaustive information at such a scale on often “forgotten” structures. In this context, the expert's role is essential, because he is able to synthesize the information resulting from the inventory and from the commonly imprecise, if not vague, criteria on the basis of his experience and his knowledge of the geological, historical, economic regional context.In this article, we propose mathematical tools for representing and processing this information in order to give flexibility to this step and manage the uncertainty inherent in the expert's information. The first tool, based on the weight of evidence theory, is for managing the uncertainty due to the heterogeneous spatial distribution of the data, whereas the second tool, based on the fuzzy set theory, is for managing the imprecision and incompleteness of available data, which hinder the definition of the class boundaries of the quantitative decision criteria. Based on an appropriate representation of the uncertainty sources (related to the input data and to the expert diagnostic), we then propose a methodology that integrates the uncertainty in the final output of the collapse susceptibility assessment and provides a confidence indicator useful within the decision-making process. The proposed methodology is applied to the Arras territory in the North of France, where abandoned chalk pits (dating back to the Roman ages) and war saps located in the vicinity of the First World War front lines (i.e. covered trenches), raise both difficulties for urban planning.