淮河流域水文极值预测模型研究

为探索气候变化影响下水文极值的非平稳性和预测方法,建立了水文极值非平稳广义极值（GEV）分布的统计预测模型。利用1952-2010年淮河上游流域累计面雨量和流量年最大值资料、同期500 hPa环流特征量资料以及17个CMIP5模式对环流特征量的模拟结果,筛选出对水文极值影响显著的年平均北半球极涡强度指数作为GEV分布参数的预测因子。分析了在RCP2.6、RCP4.5和RCP8.5情景下2006-2050年淮河上游流域水文极值对气候变化的响应。结果表明,10年以下与10年以上重现期的水文极值在非平稳过程中呈现前者下降而后者上升的相反变化趋势;多模型预测的集合平均在未来情景中均呈现上升趋势,情景排放量越大增幅越大,重现期越长增幅也越大。与极值的常态相比,极值的极端态更易受气候变化影响。     

四川盆地短历时强降水极值分布的研究

运用广义帕雷托分布(GPD)和广义极值分布(GEV),借助于L-矩的参数估计方法,对四川盆地12站的小时极端降水量进行拟合,并对两种模型的拟合效果进行比较。运用Hill图,结合统计量D*来确定GPD的最佳门限值是合适的,选出的样本是独立的。各站的小时极端降水概率分布均符合GPD和GEV,但GPD模型的拟合精度要优于GEV模型。利用两种模型推算出各站给定重现期的最大小时降水量,其中泸州50 a一遇和100 a一遇的降水极值分位数都超过了100 mm,除了遂宁站外,两种模型估计出的极值分位数的相对误差基本都在10%以下。通过分析,GPD推算的结果更加可靠。     

全球增暖1.5/2℃下中国区域极端降水的风险变化及其影响因子

利用24个CMIP6全球气候模式的逐日降水模拟资料,基于广义极值分布（GEV）模型,研究了全球增暖1.5/2℃下我国20、50和100 a重现期极端降水的未来风险变化。可以发现,相对于历史时期（1995—2014年）,全球升温1.5和2℃下极端降水发生概率风险空间分布相近,总体上呈现增加趋势,但额外增暖0.5℃将导致更高的风险。如50 a重现期极端降水,在增暖1.5/2℃下其重现期将分别变为17/14 a,极端降水将变得更加频繁。不同区域对气候变暖的响应存在区域差异,其中中国西部长江黄河中上游和青藏高原地区、中国东部长江黄河中下游及其以南地区,极端降水发生概率比达到3以上,局部更是达到5以上,为我国极端降水气候变化响应高敏感区域。进一步,基于概率分布函数从理论角度探讨了位置和尺度参数对发生概率风险的影响与贡献度量,并用于探讨极端降水气候平均态和变率变化对极端降水发生风险的影响,结果显示：位置和尺度参数的增量变化、风险变化率存在着显著的东西部差异,从而导致极端降水发生风险的影响因素存在差异。如中国西部尽管极端降水气候平均态和变率变化幅度不大,但因风险变化率较高,从而导致该区域的发生风险大幅增加;与之相反,中国东部风险变化率较小,但气候平均态和年际变率增幅较大,同样导致该区域风险增加依然较高;此外,相对于位置参数,全国大部分区域主要是尺度参数的变化导致极端降水未来风险增大。     

极值统计理论的进展及其在气候变化研究中的应用

着重论述极值统计分布在极端天气气候事件和重大工程设计中的重要意义,综述该领域国内外研究进展。例如,基于超门限峰值法（POT）的广义帕累托分布（GPD)和基于单元极大值法（BM）的广义极值分布（GEV）及其参数间的理论关系;采用极值分布模型与多状态一阶Markov链相结合构建降尺度模型模拟局地极端降水事件,推算一定重现期的极端降水量的分位数;探讨极值分布模型分位数估计误差问题,多维极值分布理论及其应用等问题。     

未来情景下南水北调中线工程水源区极端降水分布特征

利用南水北调中线工程水源区9个气象站点1961-2008年的日降水资料和IPCC第四次评估报告多模式数据结果,抽取逐年的最大日降水量序列样本,运用广义极值分布（GEV）和广义帕累托分布 (GPD）两种极值统计模型对样本进行拟合,遴选出描述流域最大日降水量分布规律的最优概率模型,推算重现期对应的降水量值,并预估该流域极端降水事件在未来气候变化情景下的响应。研究表明：南水北调中线工程水源区降水极值均符合GEV和GPD分布,但GPD模型更适合用于描述该流域降水极值分布;未来气候变化情景下用GPD分布拟合的降水极值优于使用GEV分布;A2情景下极端降水事件的发生将更频繁、更强烈,A1B情景下次之,B1情景下相对较小,表明未来高排放气候情景对极端降水事件的影响比中、低排放情景大。     

基于动态重现期的极端高温气候评估方法研究

利用全国754站逐日最高气温观测序列,在论证极端温度概率分布与非平稳性关系的基础上,构建和比较了多种非平稳广义极值模型,定义了极端高温的动态重现期和重现水平,提出了一种极端高温事件的新型评估思想和方法,并将其应用于极端气候变化研究。通过该方法可以更好地解释极端事件的真实极端性,有效地增强极端事件之间的可比性,从而保留更多历史极端气候事件的信息。动态重现期的变换运用可对当前极端事件发生的真实状态和趋势提出更准确评估。该方法的提出可有效澄清学术领域和公共舆论对于多年一遇极端事件的理解上长期混淆重现期的绝对值和概率性这一分歧和谬误。     

基于两种分布的勉县年最大日降水量重现期估算

2021年8月22日勉县遭遇极端降水事件,日降水量高达2379 mm,灾害十分严重。统计分析1959—2021年勉县历史逐年最大日降水量特点,采用皮尔逊Ⅲ型（简称P-Ⅲ型）曲线分布和耿贝尔极值分布方法推算重现期及降水量,并将两者进行比较,对2021年8月22日极端大暴雨进行重现期估算。结果表明：勉县年最大日降水年际变化明显,2008年以来变率增大且有更极端的趋势;基于P-Ⅲ型曲线分布和耿贝尔极值分布的1959—2020年最大日降水积累概率拟合效果均较好,但耿贝尔极值分布对年最大日降水量的拟合优于P-Ⅲ型分布;应用耿贝尔极值分布推算勉县极值降水,100 a一遇的日降水量为1547 mm,2021年8月22日降水量2379 mm的重现期为4 88133 a。增加2021年最大日降水量进入样本序列重新构建耿贝尔极值分布函数,推算日降水量2379 mm的重现期为70735 a,100 a一遇的估算降水量为1834 mm,重现期及降水量估算变化均较大,说明超极端降水和样本长度对重现期的推算影响较大。     

南京过去100年极端日降水量模拟研究

在南京过去100年日降水资料的基础上,利用极值理论中的区组模型和阈值模型分析了极端日降水分布特征.首先通过广义极值(GEV)模型模拟了日降水的年极值序列(AMDR),用极大似然估计(MLE)方法计算了模型的参数,并借助轮廓似然函数估计出参数的精确误差区间,同时采用4种较直观的诊断图形对模型的合理性进行全面评估,结果表明Frechet是区组模型中最适合描述极端日降水分布特征的函数.其次,将日降水序列分3种情景构建极值分布的阈值模型(GPD),考察了观测数据的规模对应用该模型的限制,重点讨论了如何针对给定观测样本选择合适的阈值收集极值信息.分析结果认为,长度不小于50年的气候序列,采用24 mm的日降水量作为临界阈值均能进行GPD分析.该阈值处于年降水序列第91个百分位附近,即对目前长度为50年左右的日观测资料,第91个百分位点以上的数据基本能满足GPD研究的需要.另外,根据GEV和GPD对未来极端降水重现水平的推断情况,GPD预测值的置信区间要比GEV的窄,极值推断的不确定性相对也较小,更适合用于研究中国目前规模不大的气候资料.最后,对GPD模型的形状参数和尺度参数进行变换,分别引入描述线性变化的动态变量,分析降水序列中潜在的变异行为对极值理论应用的影响.这种变异包括降水序列中长期的均值变化及百分位变化,从模拟结果看,暂未发现资料变异行为对极值分析产生显著于扰.     

长江三峡地区宜昌、巴东短历时极值降水特征分析

利用长江三峡库首宜昌站及库区巴东站1955—2008年分钟降水强度资料,采用广义极值分布和线性矩参数估计方法,拟合两站7个短历时(60min以内)年最大降水量概率分布,推断各历时有关重现期降水极值,计算各历时暴雨频次及年最大降水量气候倾向率,分析各历时降水广义极值分布的参数随时间变化规律。结果表明:宜昌、巴东两站7个短历时年最大降水量采用广义极值分布拟合,其效果较好;两站短历时降水平均值趋势变化不明显,而不同百分位数降水量变化趋势差异较大,其中中位数的降水量呈下降趋势,较高百分位数的降水量增加趋势显著,达20%～30%。     

广义极值分布理论在重现期计算的应用

在气候统计学上,常用Weibull、Gumbel、Frechet统计分布函数对极端气候要素的分布进行拟合,广义极值分布理论综合了以上三种极值分布模型,在气候分析中得到了广泛应用。以南昌市年汛期日最大降水量为例,利用广义极值分布理论对其分布进行拟合,并对重现值及其置信区间进行计算,为气候要素极值的统计分析提供了一种新的手段。     

Risk assessment of precipitation extremes in northern Xinjiang,China

This study was conducted using daily precipitation records gathered at 37 meteorological stations in northern Xinjiang, China, from 1961 to 2010. We used the extreme value theory model, generalized extreme value (GEV) and generalized Pareto distribution (GPD), statistical distribution function to fit outputs of precipitation extremes with different return periods to estimate risks of precipitation extremes and diagnose aridity–humidity environmental variation and corresponding spatial patterns in northern Xinjiang. Spatiotemporal patterns of daily maximum precipitation showed that aridity–humidity conditions of northern Xinjiang could be well represented by the return periods of the precipitation data. Indices of daily maximum precipitation were effective in the prediction of floods in the study area. By analyzing future projections of daily maximum precipitation (2, 5, 10, 30, 50, and 100 years), we conclude that the flood risk will gradually increase in northern Xinjiang. GEV extreme value modeling yielded the best results, proving to be extremely valuable. Through example analysis for extreme precipitation models, the GEV statistical model was superior in terms of favorable analog extreme precipitation. The GPD model calculation results reflect annual precipitation. For most of the estimated sites’ 2 and 5-year T for precipitation levels, GPD results were slightly greater than GEV results. The study found that extreme precipitation reaching a certain limit value level will cause a flood disaster. Therefore, predicting future extreme precipitation may aid warnings of flood disaster. A suitable policy concerning effective water resource management is thus urgently required.     

中国格点化日降水极值统计模型及阈值的选取

采用年最大值法（AM）及超阈值峰量法（POT）分别构建基于0.5°×0.5°网格的全国地面日降水极值序列,建立基于广义极值分布（GEV）和广义帕累托分布（GPD）的降水极值统计模型,通过K-S检验评估模型拟合效果,研究全国日降水极值的统计规律及其空间分布特征,提出适用于不同地区极端日降水的极值分布模型与阈值选取标准,结果表明：(1)POT序列比AM序列更符合降水极值序列的要求;(2)为便于比较并提高模型拟合效果,POT序列的阈值由百分位数法确定效果较好;(3)阈值方案优选结果在空间分布上与中国干湿区域的划分有很好的相关性,在湿润地区宜将第90～94百分位数作为阈值,在半湿润和半干旱地区宜将第94～97百分位数作为阈值,在干旱地区则使用第97～99百分位数较为合适。     

Temperature Extremes from Canadian Regional Climate Model (CRCM) Climate Change Projections

Future climate projections of extreme events can help forewarn society of high-impact events and allow the development of better adaptation strategies. In this study a non-stationary model for Generalized Extreme Value (GEV) distributions is used to analyze the trend in extreme temperatures in the context of a changing climate and compare it with the trend in average temperatures.

The analysis is performed using the climate projections of the Canadian Regional Climate Model (CRCM), under an IPCC SRES A2 greenhouse gas emissions scenario, over North America. Annual extremes in daily minimum and maximum temperatures are analyzed. Significant positive trends for the location parameter of the GEV distribution are found, indicating an expected increase in extreme temperature values. The scale parameter of the GEV distribution, on the other hand, reveals a decrease in the variability of temperature extremes in some continental regions. Trends in the annual minimum and maximum temperatures are compared with trends in average winter and summer temperatures, respectively. In some regions, extreme temperatures exhibit a significantly larger increase than the seasonal average temperatures.

The CRCM projections are compared with those of its driving model and framed in the context of the Coupled Model Intercomparison Project, phase 3 (CMIP3) Global Climate Model projections. This enables us to establish the CRCM position within the CMIP3 climate projection uncertainty range. The CRCM is validated against the HadEX2 dataset in order to assess the CRCM representation of temperature extremes in the present climate. The validation is also framed in the context of CMIP3 validation results. The CRCM cold extremes validate better and are closer to the driving model and CMIP3 projections than the hot extremes.     

Non-stationary extreme value analysis in a changing climate

This paper introduces a framework for estimating stationary and non-stationary return levels, return periods, and risks of climatic extremes using Bayesian inference. This framework is implemented in the Non-stationary Extreme Value Analysis (NEVA) software package, explicitly designed to facilitate analysis of extremes in the geosciences. In a Bayesian approach, NEVA estimates the extreme value parameters with a Differential Evolution Markov Chain (DE-MC) approach for global optimization over the parameter space. NEVA includes posterior probability intervals (uncertainty bounds) of estimated return levels through Bayesian inference, with its inherent advantages in uncertainty quantification. The software presents the results of non-stationary extreme value analysis using various exceedance probability methods. We evaluate both stationary and non-stationary components of the package for a case study consisting of annual temperature maxima for a gridded global temperature dataset. The results show that NEVA can reliably describe extremes and their return levels.     

Observed Changes in One-in-20 Year Extremes of Canadian Surface Air Temperatures

Weather and climate extremes are often associated with substantial adverse impacts on society and the environment. Assessment of changes in extremes is of great and broad interest. This study first homogenizes daily minimum and maximum surface air temperatures recorded at 146 stations in Canada. In order to assess changes in one-in-20 year extremes (i.e., extremes with a 20-year return period) in temperature, annual maxima and minima of both daily minimum temperatures and daily maximum temperatures are derived from the homogenized daily temperature series and analyzed with a recently developed extreme value analysis approach based on a tree of generalized extreme value distributions (including stationary and non-stationary cases). The procedure is applied to estimate the changes over the period 1911 to 2010 at 115 stations, located mainly in southern Canada, and over the period 1961 to 2010 at 146 stations across Canada (including 37 stations in the North). The results show that warming is strongest for extreme low temperature and weakest for extreme high temperature and is much stronger in the Canadian Arctic than in southern Canada. Warming is stronger in winter than in summer and stronger during nighttime than daytime of the same season.