元江—红河流域水文过程受山地气候和人类活动影响,时空分异性明显;近半个世纪以来,其河川径流变化日趋复杂,跨境水安全风险日益凸显.基于流域55a逐月径流、降水和气温站点数据,利用GAMLSS(Generalized Additive Models for Location,Sc...元江—红河流域水文过程受山地气候和人类活动影响,时空分异性明显;近半个世纪以来,其河川径流变化日趋复杂,跨境水安全风险日益凸显.基于流域55a逐月径流、降水和气温站点数据,利用GAMLSS(Generalized Additive Models for Location,Scale and Shape)建立以时间、气候因子为协变量的时变矩模型,系统分析了综合环境影响下径流序列的非一致性规律及其对气候变化的响应;在非一致性框架下基于重现期的期望超过次数法(Expected Number of Exceedances,ENE)推求设计年径流,探讨了非一致性对流域水资源利用与管理的潜在影响.结果表明:(1)流域径流具有显著非一致性特征,以哀牢山系为界,以东的干游元江倾向于突变性且存在一个位于2002年的显著突变点,以西的支流李仙江在1986年转折点后的局部趋势呈现显著下降;(2)气候因子为协变量的GAMLSS模型分别服从伽马和对数正态分布时,为径流序列提供了最佳拟合,可有效地捕捉气候变化对元江和李仙江径流的影响;(3)与传统一致性模型相比,时变矩模型为计算变化环境下的设计年径流提供了更科学的方法.在选定的排放情景下,基于时变矩模型得到的元江和李仙江设计年径流值较一致性模型,分别表现为-15.12~25.46%和-17.29~22.24%的差异变化,这些差异对水资源开发规模、动态管理和利用过程具有重大影响.研究成果可为元江—红河流域跨境水资源的合理利用与协调管理提供新的科学依据.展开更多
为探讨水文数据非一致性对洪水频率分析的影响,提出基于位置、尺度、形状的广义可加模型(Generalized additive models for location,scale and shape,GAMLSS),从时间和降水两类因素出发,计算单变量、多变量洪水频率,分析经验点据与理...为探讨水文数据非一致性对洪水频率分析的影响,提出基于位置、尺度、形状的广义可加模型(Generalized additive models for location,scale and shape,GAMLSS),从时间和降水两类因素出发,计算单变量、多变量洪水频率,分析经验点据与理论分位曲线拟合效果,选取不同变化条件下洪水适宜理论分布。以汤旺河流域为例,研究结果表明,单变量洪水频率最优分布选取较稳定,而受时间和降水因素影响,多变量洪水频率最优分布选取均不同。与前者相比,引入协变量使原序列参考时间连续性变化和降水极端信息,改进传统洪水频率计算方法。展开更多
The UK’s economic growth has witnessed instability over these years. While some sectors recorded positive performances, some recorded negative performances, and these unstable economic performances led to technical r...The UK’s economic growth has witnessed instability over these years. While some sectors recorded positive performances, some recorded negative performances, and these unstable economic performances led to technical recession for the third and fourth quarters of the year 2023. This study assessed the efficacy of the Generalised Additive Model for Location, Scale and Shape (GAMLSS) as a flexible distributional regression with smoothing additive terms in forecasting the UK economic growth in-sample and out-of-sample over the conventional Autoregressive Distributed Lag (ARDL) and Error Correction Model (ECM). The aim was to investigate the effectiveness and efficiency of GAMLSS models using a machine learning framework over the conventional time series econometric models by a rolling window. It is quantitative research which adopts a dataset obtained from the Office for National Statistics, covering 105 monthly observations of major economic indicators in the UK from January 2015 to September 2023. It consists of eleven variables, which include economic growth (Econ), consumer price index (CPI), inflation (Infl), manufacturing (Manuf), electricity and gas (ElGas), construction (Const), industries (Ind), wholesale and retail (WRet), real estate (REst), education (Edu) and health (Health). All computations and graphics in this study are obtained using R software version 4.4.1. The study revealed that GAMLSS models demonstrate superior outperformance in forecast accuracy over the ARDL and ECM models. Unlike other models used in the literature, the GAMLSS models were able to forecast both the future economic growth and the future distribution of the growth, thereby contributing to the empirical literature. The study identified manufacturing, electricity and gas, construction, industries, wholesale and retail, real estate, education, and health as key drivers of UK economic growth.展开更多
Recent years have witnessed increasingly frequent extreme precipitation events,especially in desert steppes in the semi-arid and arid transition zone.Focusing on a desert steppe in western-central Inner Mongolia Auton...Recent years have witnessed increasingly frequent extreme precipitation events,especially in desert steppes in the semi-arid and arid transition zone.Focusing on a desert steppe in western-central Inner Mongolia Autonomous Region,China,this study aimed to determine the principle time-varying pattern of extreme precipitation and its dominant climate forcings during the period 1988-2017.Based on the generalized additive models for location,scale,and shape(GAMLSS)modeling framework,we developed the best time-dependent models for the extreme precipitation series at nine stations,as well as the optimized non-stationary models with large-scale climate indices(including the North Atlantic Oscillation(NAO),Atlantic Multidecadal Oscillation(AMO),Southern Oscillation(SO),Pacific Decadal Oscillation(PDO),Arctic Oscillation(AO),and North Pacific Oscillation(NPO))as covariates.The results indicated that extreme precipitation remained stationary at more than half of the stations(Hailisu,Wuyuan,Dengkou,Hanggin Rear Banner,Urad Front Banner,and Yikewusu),while linear and non-linear time-varying patterns were quantitatively identified at the other stations(Urad Middle Banner,Linhe,and Wuhai).These non-stationary behaviors of extreme precipitation were mainly reflected in the mean value of extreme precipitation.The optimized non-stationary models performed best,indicating the significant influences of large-scale climate indices on extreme precipitation.In particular,the NAO,NPO,SO,and AMO remained as covariates and significantly influenced the variations in the extreme precipitation regime.Our findings have important reference significance for gaining an in-depth understanding of the driving mechanism of the non-stationary behavior of extreme precipitation and enable advanced predictions of rainstorm risks.展开更多
采用北江流域10个水文站点1959-2005年日流量数据,运用滑动秩和检验法(Mann-Whitney U test)和AMOC检验法确定样本最佳突变点,采用Mann-Kendall(MK)检验法检测时间趋势性,最后结合广义可加模型(GAMLSS)分别构建洪水量级和频率与影响因...采用北江流域10个水文站点1959-2005年日流量数据,运用滑动秩和检验法(Mann-Whitney U test)和AMOC检验法确定样本最佳突变点,采用Mann-Kendall(MK)检验法检测时间趋势性,最后结合广义可加模型(GAMLSS)分别构建洪水量级和频率与影响因子的非平稳性模型.研究表明:1)北江流域年及季节洪峰流量普遍呈下降或显著下降趋势,在1990年发生突变,但年最大洪峰无显著变化趋势;2)1990年之后,最大三场洪水及重现期大于10年的洪水事件多集中发生,且洪水发生频率、量级及峰现时间均发生较大改变;3)LOGNO分布为年最大洪峰流量序列最优极值分布,计算出非平稳性条件下北江流域4个站点百年一遇设计洪水流量值,并分别以时间(Model 1)和气候指标(Model 2)为解释变量对洪水发生次数进行模拟分析,充分反映洪水发生次数的随机过程,为区域防洪减灾提供理论依据.展开更多
文摘元江—红河流域水文过程受山地气候和人类活动影响,时空分异性明显;近半个世纪以来,其河川径流变化日趋复杂,跨境水安全风险日益凸显.基于流域55a逐月径流、降水和气温站点数据,利用GAMLSS(Generalized Additive Models for Location,Scale and Shape)建立以时间、气候因子为协变量的时变矩模型,系统分析了综合环境影响下径流序列的非一致性规律及其对气候变化的响应;在非一致性框架下基于重现期的期望超过次数法(Expected Number of Exceedances,ENE)推求设计年径流,探讨了非一致性对流域水资源利用与管理的潜在影响.结果表明:(1)流域径流具有显著非一致性特征,以哀牢山系为界,以东的干游元江倾向于突变性且存在一个位于2002年的显著突变点,以西的支流李仙江在1986年转折点后的局部趋势呈现显著下降;(2)气候因子为协变量的GAMLSS模型分别服从伽马和对数正态分布时,为径流序列提供了最佳拟合,可有效地捕捉气候变化对元江和李仙江径流的影响;(3)与传统一致性模型相比,时变矩模型为计算变化环境下的设计年径流提供了更科学的方法.在选定的排放情景下,基于时变矩模型得到的元江和李仙江设计年径流值较一致性模型,分别表现为-15.12~25.46%和-17.29~22.24%的差异变化,这些差异对水资源开发规模、动态管理和利用过程具有重大影响.研究成果可为元江—红河流域跨境水资源的合理利用与协调管理提供新的科学依据.
文摘为探讨水文数据非一致性对洪水频率分析的影响,提出基于位置、尺度、形状的广义可加模型(Generalized additive models for location,scale and shape,GAMLSS),从时间和降水两类因素出发,计算单变量、多变量洪水频率,分析经验点据与理论分位曲线拟合效果,选取不同变化条件下洪水适宜理论分布。以汤旺河流域为例,研究结果表明,单变量洪水频率最优分布选取较稳定,而受时间和降水因素影响,多变量洪水频率最优分布选取均不同。与前者相比,引入协变量使原序列参考时间连续性变化和降水极端信息,改进传统洪水频率计算方法。
文摘The UK’s economic growth has witnessed instability over these years. While some sectors recorded positive performances, some recorded negative performances, and these unstable economic performances led to technical recession for the third and fourth quarters of the year 2023. This study assessed the efficacy of the Generalised Additive Model for Location, Scale and Shape (GAMLSS) as a flexible distributional regression with smoothing additive terms in forecasting the UK economic growth in-sample and out-of-sample over the conventional Autoregressive Distributed Lag (ARDL) and Error Correction Model (ECM). The aim was to investigate the effectiveness and efficiency of GAMLSS models using a machine learning framework over the conventional time series econometric models by a rolling window. It is quantitative research which adopts a dataset obtained from the Office for National Statistics, covering 105 monthly observations of major economic indicators in the UK from January 2015 to September 2023. It consists of eleven variables, which include economic growth (Econ), consumer price index (CPI), inflation (Infl), manufacturing (Manuf), electricity and gas (ElGas), construction (Const), industries (Ind), wholesale and retail (WRet), real estate (REst), education (Edu) and health (Health). All computations and graphics in this study are obtained using R software version 4.4.1. The study revealed that GAMLSS models demonstrate superior outperformance in forecast accuracy over the ARDL and ECM models. Unlike other models used in the literature, the GAMLSS models were able to forecast both the future economic growth and the future distribution of the growth, thereby contributing to the empirical literature. The study identified manufacturing, electricity and gas, construction, industries, wholesale and retail, real estate, education, and health as key drivers of UK economic growth.
基金funded by the Yinshanbeilu Grassland Eco-hydrology National Observation and Research Station,China Institute of Water Resources and Hydropower Research(YSS202105)the National Natural Science Foundation of China(52269005)+3 种基金the Inner Mongolia Science and Technology Plan Project(2022YFSH0105)the Central Guidance for Local Science and Technology Development Fund Projects(2024ZY0002)the Inner Mongolia Autonomous Region University Youth Science and Technology Talent Project(NJYT 22037)the Inner Mongolia Agricultural University Young Teachers'Scientific Research Ability Improvement Project(BR220104).
文摘Recent years have witnessed increasingly frequent extreme precipitation events,especially in desert steppes in the semi-arid and arid transition zone.Focusing on a desert steppe in western-central Inner Mongolia Autonomous Region,China,this study aimed to determine the principle time-varying pattern of extreme precipitation and its dominant climate forcings during the period 1988-2017.Based on the generalized additive models for location,scale,and shape(GAMLSS)modeling framework,we developed the best time-dependent models for the extreme precipitation series at nine stations,as well as the optimized non-stationary models with large-scale climate indices(including the North Atlantic Oscillation(NAO),Atlantic Multidecadal Oscillation(AMO),Southern Oscillation(SO),Pacific Decadal Oscillation(PDO),Arctic Oscillation(AO),and North Pacific Oscillation(NPO))as covariates.The results indicated that extreme precipitation remained stationary at more than half of the stations(Hailisu,Wuyuan,Dengkou,Hanggin Rear Banner,Urad Front Banner,and Yikewusu),while linear and non-linear time-varying patterns were quantitatively identified at the other stations(Urad Middle Banner,Linhe,and Wuhai).These non-stationary behaviors of extreme precipitation were mainly reflected in the mean value of extreme precipitation.The optimized non-stationary models performed best,indicating the significant influences of large-scale climate indices on extreme precipitation.In particular,the NAO,NPO,SO,and AMO remained as covariates and significantly influenced the variations in the extreme precipitation regime.Our findings have important reference significance for gaining an in-depth understanding of the driving mechanism of the non-stationary behavior of extreme precipitation and enable advanced predictions of rainstorm risks.
文摘采用北江流域10个水文站点1959-2005年日流量数据,运用滑动秩和检验法(Mann-Whitney U test)和AMOC检验法确定样本最佳突变点,采用Mann-Kendall(MK)检验法检测时间趋势性,最后结合广义可加模型(GAMLSS)分别构建洪水量级和频率与影响因子的非平稳性模型.研究表明:1)北江流域年及季节洪峰流量普遍呈下降或显著下降趋势,在1990年发生突变,但年最大洪峰无显著变化趋势;2)1990年之后,最大三场洪水及重现期大于10年的洪水事件多集中发生,且洪水发生频率、量级及峰现时间均发生较大改变;3)LOGNO分布为年最大洪峰流量序列最优极值分布,计算出非平稳性条件下北江流域4个站点百年一遇设计洪水流量值,并分别以时间(Model 1)和气候指标(Model 2)为解释变量对洪水发生次数进行模拟分析,充分反映洪水发生次数的随机过程,为区域防洪减灾提供理论依据.
