长江三角洲太湖流域属于典型的北亚热带季风气候区,气候变化、快速城镇化和人类活动使得极端气象灾害的影响加剧.极端降水是形成洪水的主要来源之一,分析降水极值特性是研究洪水的一种必要且有效的手段.以湖西浙西区96个站点20062012年的日降水数据为例,选取极端降水量(R95P)、一日最大降水量(RX1D)、最大连续五日降水量(RX5D)、降水强度(SDII)和强降水日数(R10,R20和R50)七个极端降水指数,研究降水极值空间分布规律.结果表明,极端降水量在湖西区北部呈现较高值,降水强度在浙西山区和湖西山区以及东北部地区呈现较高值,其余五个指数均显示,地势较高的浙西山区极端降水指数相应较高.进一步对丹阳、金坛、溧阳、宜兴、安吉和湖州六个代表站19612015年的降水极值分析,利用广义极值分布(Generalized Extreme Value Distribution,GEV)和广义帕累托分布(Generalized Pareto Distribution,GPD)两种极值模型进行分布拟合.研究发现,GPD的拟合效果优于GEV,各个站点的分布模型参数也有差异,对两种分布下不同重现期的极端降水量值进行对比,为风险防范和工程设计安全提供了科学依据.另外,汛期降水的空间分布差异性较大,汛期和非汛期的对比结果显示极值统计模型参数也有较大差异,主要体现在阈值和尺度参数上.对于特定季节尤其是暴雨多发期的极端降水事件,建议利用汛期的降水极值进行统计模拟,有助于更好地刻画降水极值特性.
英文摘要:
The Taihu Lake basin in the Yangtze River Delta is a typical northern subtropical monsoon climate area. Climate change, rapid urbanization and human activities aggravate the impact of extreme meteorological disasters. Extreme rainfall is one of the main sources of flood disasters. Analyzing the characteristics of extreme rainfall is a necessary and effective method to study on flood event. Taking the daily precipitation data of 96 stations from 2006 to 2012 in the western area of Taihu Lake basin as an example,seven extreme precipitation indices (R95P,RX1D,RX5D, SDII,R10,R20 and R50) were selected to study the spatial distribution of extreme precipitation. The results show that the extreme precipitation in the northern part of western area of Taihu Lake basin presents a higher value, while the other six indices show that the extreme precipitation index in mountainous areas with higher terrain is correspondingly higher. Furthermore, the precipitation extreme values of six representative stations (Danyang, Jintan, Liyang, Yixing, Anji and Huzhou) from 1961 to 2015 were analyzed, and the distribution was fitted by two extreme value statistical models,Generalized Extreme Value Distribution (GEV) and Generalized Pareto Distribution (GPD). It is found that the fitting performance of GPD is better than that of GEV, and the distribution model parameters of each station are also different. Precipitation extreme values in different return periods under two distribution models also provide guidance for risk prevention and engineering design safety. In addition,the spatial distribution of precipitation in flood season is more different. The comparison between flood season and non-flood season shows that the parameters of extreme value statistical model are also quite different, mainly reflected in the threshold and scale parameters. For extreme rainfall events in specific seasons,especially in rainstorm-prone periods,it is suggested that the extreme rainfall observations in flood season be used for statistical simulation, which is helpful to better characterize the statistical characteristics of extreme rainfall.