基于RHtest方法对新疆北部风区近50 a风速资料的均一性研究  

Homogeneity of wind speed data of recent 50 years based on the RHtest method for wind areas of northern Xinjiang

作　　者：周昊楠[1,2] 何清[1,3] 秦榕[2] 刘卫平[2] 

机构地区：[1]南京信息工程大学,江苏南京210044 [2]新疆气象信息中心,新疆乌鲁木齐830002 [3]中国气象局乌鲁木齐沙漠气象研究所,新疆乌鲁木齐830002

出　　处：《干旱区地理》2016年第3期530-538,共9页Arid Land Geography

基　　金：公益性行业(气象)科研专项(GYHY201006012)

摘　　要：利用RHtest提供的两种检验方法:惩罚最大T检验(The Penized Maximal T Test,PMTT)和惩罚最大F检验(The Penized Maximal F Test,PMFT),对新疆大风区域13个国家级气象站点建站至2014年逐年平均风速资料进行了均一性检验和订正;通过在新疆地区特定环境下对该两种误报率分布相对均匀且微小偏移检测能力较高的方法进行评估,找出每种方法在新疆地区实际应用当中的适用性,最终通过订正改善新疆风要素年值历史序列的均一性,结果表明:(1)站址迁移、仪器换型、测风手段变化及台站环境变化等非气候变化因素均对13个待检站点年平均风速序列的均一性产生显著影响,其中仪器换型造成的影响尤为突出。(2)当参考站均一性较好时,PMTT方法检测结果与台站元数据一致性较好,反之则存在虚假跳变点或漏检。(3)PMFT方法因为不依靠参考站,能很好的弥补新疆大风区域站点稀疏、参考站点筛选困难和元数据不完整等弊端,该方法对迁站和仪器换型的检测能力较强。(4)数据订正后,总体变化趋势没有大的改变,呈现下降趋势,符合风要素在全国范围的总体变化趋势,数据的均一性得到很大改善。Two tests of RHtest method, the Penalized Maximal T Test （PMTT） and the Penalized Maximal F Test （PMFT）, were used in this paper to conduct homogeneity test and data correction for the annual average wind speed data of 13 national meteorological stations in gale areas of Xinjiang, China from their establishment to 2014. The introduction of empirical penalty function has greatly improved the non-uniform distribution of false alarm rate and the algorithms test capability. PMTT is a test method that depends on reference sequence and uses the average offset degrees of the time buckets before and after different nodes in the sequence to find the discontinuity point. Due to the consideration of the relative position among different nodes, this method can eliminate the impact of different sample lengths on test results. While PMFT does not require a reference sequence in the test, but was proposed based on a series improvement of the two-phase regression statistical test, which can avoid effectively the test error due to non-uniform reference sequence. In this paper, comparative analysis was conducted between the results of two test methods, and finally created the corrected data sequence. Results show as follows：（1） the factors of non-climate change such as station relocation, instrument model change, anemometry means change and station environment change had big impacts on the homogeneity of annual average wind speed sequences of the 13 stations in the wind areas of Xinjiang, of which the instrument model change had the biggest impact.（2） According to PMFT, all the 13 stations had significant trip points at the significance level of 95%;while for PMTT method, only 10 stations met the testing requirements under the constraints of reference station distribution and factor dependency, and even appeared false trip points.（3） When the reference station was of good homogeneity, the test results of PMTT method was in good consistency with the station metadata, and showed good test effect for subtle chan

关 键 词：大风区域 风速 均一性 检验 订正 

分 类 号：P412.16[天文地球—大气科学及气象学]