运用EOF分解、线性趋势分析、方差分析、小波分析、相关分析等方法,从均值、长期变化、年际和年代际变化特征等方面对NCEP/NCAR(N1T)、NCEP/DOE(N2T)、ERA—Interim(ET)、JRA-25(JT)4种再分析资料集和194站观测资料集中1979-2012年我国冬季最低气温(WTMIN)进行比较。结果表明:(1)ET在再现观测冬季最低气温多年平均时最优;(2)除新疆北部和西南部以及东北北部有降低趋势外,我国大部分地区WTMIN有升高的趋势,N1T对观测温度长期变化趋势再现能力优于其他3种资料;(3)观测WTMIN在中国北方和高原地区年际和年代际变化大于南方地区,四种再分析WTMIN对我国冬季最低气温的年际和年代际变化的再现在不同区域各有所长,总体上以ET为优。(4)再分析资料在再现观测WTMIN年代际时间尺度上的能力相当,以ET最优,尤其是在我国中东部地区和新疆北部地区。NIT和N2T对东北地区观测冬季最低气温年际变化的再现能力较好,ET再现观测WTMIN在1990s之前,2005年之后的3~4a变化周期特征时能力略优于其他3种资料。
By using EOF decomposition, linear trend analysis, variance analysis, wavelet analysis, correlation coefficient method, the winter minimum temperature observed from datasets during 1979-2012 obtained from 194 stations in China were compared with that from NCEP/NCAR (N1T),NCEP/DOE (N2T) ,ERA-Interim (ET) and JRA-25 (JT) reanalysis datasets in terms of the average, the long term variation, the interannual and the interdecadal variations. Results suggest that : ( 1 ) the ET performs best in reproducing the observed multi-year average winter minimum temperature. (2) Except the northern and southeastern Xinjiang and the north of Northeast China, the winter minimum temperature in most parts of China increased during 1979-2012; NIT performs better than other three datasets in depicting the linear trend of the observed temperature. ( 3 ) Annual and interdecadal variations of observed winter minimum temperature are larger in northern China and Tibetan plateau than those in southern China; the four reanalysis of winter minimum temperature show different depiction in different area, generally the ET is better. (4) The four reanalysis data are equal in reproducing the interdecadal time scale features of ob- served winter minimum temperature, and ET is better, especially in middle and east of China and the northern Xinjiang. N1T and N2T have better ability in representing the annual variability of observed win- ter minimum temperature in northeast of China,while ET performs better in representing the features of 3 -4 periods before 1990 and after 2005 than the other three reanalysis data.