摘要
自然界中的降雨由于受到季风、台风等影响会表现出非季节性,因而不符合常规极值模型的假定。因此通过似然比检验,本文采取季度模型,将降雨分分组为四个季度,采用广义极值分布来进行月最大降雨量的模拟。另外采用经验度来进行最大重现期的估计,即多大程度的外推时段内是合理的。应用结果表明:基于90个月的降雨数据,夏季的降雨近似服从Gumbel分布,计算中存在一定的误差,而春、秋、冬最大的外推时段分别为54.2、68.6、84.6个月。该理论方法具有较高的可靠性与实用性。
In the context of environmental processes,non-stationarity of precipitation is often apparent because of seasonal effects like monsoon,typhoon,etc.Because non-stationarity violates the assump-tion of extreme models,this paper adopts the seasonal models that divide the rainfall data into four seasons and are examined by likelihood test.GEV is employed to analyze the monthly maximum rainfall.In addition,degree of experience is used to estimate the maximum return period,that is,how large the extrapolation limit is.The result shows that based on the 90-month rainfall data,the summer rainfall is subject to the Gumbel distribution with an inaccurate extrapolation,while the maximum extrapolation periods in spring,autumn and winter are 54.2,68.6 and 84.6 months respectively.Therefore,it is recommended that degree of experience could be applied to engineering because of its reliability and practicability.
出处
《气候变化研究快报》
2019年第2期160-167,共8页
Climate Change Research Letters
