【作者】 杨青； 刘新春； 霍文； 郑伟；

【Author】 Yang Qing1, Liu Xinchun1, Huo Wen1, Zheng Wei2 (1.Institute of Desert Meteorology, CMA, Urumqi 830002, China; 2. Tazhong Weather Station, Qiemo 841000, Xinjiang, China)

【机构】 中国气象局乌鲁木齐沙漠气象研究所； 塔中气象站；

【摘要】 利用塔里木盆地周边27个气象站1961—2006年逐月平均气温和塔中气象站1999-2006年逐月平均气温资料,同时选取1961—2006年NCEP/NCAR2.5°×2.5°地表2m的月平均气温再分析格点资料,通过相关分析和EOF,分别用3种方法对塔中气象站1961—1998年历年逐月平均气温序列进行了恢复与重建,分析了误差,并与周边气象站的变化特征进行对比。结果表明,逐步回归和EOF法都能够作为重建塔中逐月平均气温的方法,但相对而言,逐步回归法重建的序列误差更小,平均拟合绝对误差为0.30℃,最大绝对误差为1.89℃。而NCEP/NCAR资料由于冬季存在明显的系统性误差,数值显著偏高,不能用于塔中气温序列的重建。更多还原

【Abstract】 In this paper, with the relevant analysis and EOF methods, reconstructed series of monthly average temperature from 1961 to 1998 in middle Taklimakan Desert weather station base the monthly temperature of 27 weather stations around Tarim basin from 1961 to 2006 and the data of middle Taklimakan desert weather station from 1999 to 2006, the same time, based on the grid data of NCEP/NCAR 2.5°×2.5° on the surface monthly average temperature from 1961 to 2006.Analyzed the error, and compared the change character with the neighboring weather stations, the results are as follows: the methods of stepwise regression and EOF may used to reconstruct series of monthly average temperature of middle Taklimakan desert weather station, but the error of methods of stepwise regression is relatively little, and the average absolute error is 0.30℃ , the largest absolute error is 1.89℃. Since the NCEP/NCAR data of winter there is an obvious systematic error, and the value is significantly higher, can not be used for the temperature reconstruction of middle Taklimakan desert weather station.更多还原