摘要
目前,热带气旋预报性能的检验和分析多采用各中心每年台汛后整编的最佳路径数据集(即"年鉴")资料作为真值。然而,由于年鉴资料通常在次年才能发布,所以在业务上,常以实时定位、定强资料作为"真值"进行预报性能的检验,因而不同机构(口径)给出的预报性能往往不尽相同,造成了混乱。此外,实际业务预报中,因没有实时的年鉴资料,各预报方法的起报位置只能采用实时业务定位,显然不可避免地导致了误差。为分析使用实时定位和年鉴作为"真值"进行预报性能检验的差异、评估定位误差对预报性能造成的可能影响,本文首先考察最佳路径和实时/初始定位之间的差异(即定位误差)及其分布特征,然后分析采用实时/初始定位和最佳路径作为"真值"计算预报误差时的差异,最后基于最基础的气候可持续性(Climatology and Persistence,CLIPER)预报方法初步评估了预报性能对定位误差的敏感性。结果表明:以中国气象局整编的年鉴(CMA-STI的最佳路径数据集)资料为"真值",2013—2019年间国内外各主要预报机构及全球模式的定位误差平均为24.3 km;若以东京台风中心(RSMC-Tokyo)的年鉴资料为"真值",则定位误差平均为26.2 km。分析发现,定位误差与强度密切相关,热带风暴阶段的定位误差高达35.7~41.1 km,而超强台风阶段的定位误差仅为7.5~8.3 km;在96 h预报时效内,以最佳路径为"真值"计算得到的平均预报误差均略小于以实时/初始定位为"真值"的误差,但强度越强差异越小;定位误差对短时效内的预报性能有较显著的影响。
In so far,the post-seasonal best track datasets given by each regional center are used as the true values of tropical cyclone forecast performance verification and analysis.Operationally,the verification regards real-time positioning as the true values as a result of the best track datasets published after one year,which leads to an obvious inevitable error.Consequently,different organizations give out distinct forecast performance,which causes confusions.In order to analyze the bias in forecast performance verifications by using the real-time/initial positioning and the best track datasets as the“true values”respectively and assess the possible effect of positioning error on forecast performance,this paper firstly investigates the bias between the best track datasets and the real-time/initial positioning(i.e.the positioning error)and its distribution characteristics,then analyzes the difference when using the real-time/initial positioning and the best track datasets as the“true values”to calculate the forecast errors respectively,and finally preliminarily assesses the sensitivity of forecast performance to positioning error based on the most basic CLIPER(Climatology and Persistence)forecast method.Results show that when the best track datasets from CMA-STI are taken as the“true values”,the average positioning error is 24.3 km from 2013 to 2019,and when the datasets from RSMC-Tokyo are taken as the“true values”,the error is 26.2 km.It found that the positioning error is closely related to the intensity.The positioning error in tropical storm stage is 35.7-41.1 km,while that in super typhoon stage is only 7.5-8.3 km.Within 96 h lead time,the average forecast error calculated with the best track datasets as the“true values”is slightly less than that calculated with the real-time/initial positioning as the“true values”,but the stronger the intensity,the smaller the difference.The positioning error has a significant impact on the forecast performance in a short lead time.
作者
褚萌
雷小途
陈国民
CHU Meng;LEI Xiaotu;CHEN Guomin(Chinese Academy of Meteorological Sciences,Beijing 100081,China;Shanghai Typhoon Institute,China Meteorological Administration,Shanghai 200030,China;Shanghai Meteorological Bureau,Shanghai 200030,China)
出处
《大气科学学报》
CSCD
北大核心
2022年第1期22-29,共8页
Transactions of Atmospheric Sciences
基金
国家重点研发计划项目(2018YFC1506400)。
关键词
热带气旋
定位误差
预报误差
最佳路径
tropical cyclone
positioning error
forecast error
best track
