中国多尺度不同量级极端降水发生率非平稳性研究

利用中国1951—2014年728个气象站点日降水数据,采用核估计技术、Cox回归模型、泊松回归和广义可加模型(GAMLSS)等全面分析了不同阈值条件下基于超阈值(POT)抽样的中国极端降水发生率非平稳性特征。研究表明:(1)西北部极端降水发生率在年际上分布最不均匀,阈值的增加导致不均匀程度和范围加深和扩展。西北部和东南部年际尺度极端降水发生率呈显著上升趋势,可能引发更严重的洪涝灾害;中部和东北部则相反,极端降水频率趋于减弱;(2)基于Cox回归模型的分析表明,厄尔尼诺/南方涛动(SOI)、北大西洋涛动(NAO)、印度洋偶极子(IOD)和太平洋年代际涛动(PDO)等气候指标均为影响不同区域年内尺度极端降水发生率的显著气候因子,年内尺度极端降水发生率在很大程度上依赖气候指标的变化,呈现出非平稳性特征;(3)除西北部外,其他大部分区域极端降水年发生次数没有展现过于离散的特征,然而阈值的增加导致出现过于离散特征的倾向愈益明显。当太平洋年代际涛动(PDO)和厄尔尼诺/南方涛动(SOI)处于正相位且越大时,中国西部和中部年极端降水发生次数将随之增加,东北部则相反。气候指标与850 hPa的回归关系表明风向及携带的水汽量则可能是气候指标影响极端降水频率的因素。

The spatiotemporal rates of heavy precipitation occurrence at difference scales in China

The spatiotemporal rates of heavy precipitation occurrence were analyzed comprehensively based on Peak-over-Threshold (POT) resampling under different thresholds by kernel estimating method, Poisson regression model and GAMLSS according to daily precipitation data of 728 stations during 1951-2014 over China. The results indicate that:(1) the time of heavy precipitation occurrence has the most un-uniform in inter-annual on northwest arid areas. In addition, the degree and scope of uneven is increasing and ex-panding with bigger threshold. The rate of heavy precipitation occurrence is significantly increasing in in-ter-annual in northwestern and southeastern China, which may cause more severe flooding; central and northeastern China are on the contrary, which may have a higher drought risk.(2) The rate of heavy pre-cipitation occurrence of different areas in inner-year are influenced by different climate indices based on Cox regression model, which indicates that the occurrence of heavy precipitation events are not independent but dependent on the changes of the climate indices.(3) Almost all areas didn''t exhibit over-dispersed in annual heavy precipitation occurrence times except northwest areas. However, annual heavy precipitation oc-currence time trends to temporal over-dispersed with threshold increasing. The occurrence times of heavy oc-currence are increasing with Southern oscillation index (SOI) and Pacific Decadal Oscillation (PDO) in-creasing in most areas except northeast areas.