ERA-Interim再分析和NCEP FNL分析资料在东南极中山站至Dome A断面的适用性研究

A SURFACE CLIMATOLOGICAL VALIDATION OF ECMWF ERA-INTERIM REANALYSIS AND NCEP FNL ANALYSIS OVER EAST ANTARCTICA

利用2008年南极中山站至Dome A断面上观测站的近地面气象观测资料对ECMWF ERA-Interim再分析和NCEP FNL分析资料在东南极地区的适用性进行了验证。结果表明,ERA-Interim再分析资料的气温表现明显优于FNL分析资料,其与观测的年均绝对偏差在南极大陆沿岸地区<1℃,在内陆高原<2℃;而FNL分析资料的气温在南极内陆高原地区较观测明显偏暖,尤其在冬季偏暖达8—10℃,表明其不能直接用于南极内陆高原气温的变化分析。FNL的地面气压与观测比较接近,逐月平均偏差仅约1 h Pa,其精度明显高于ERA-Interim再分析资料的地面气压,而后者在沿岸地区存在明显的系统性偏低。ERA-Interim和FNL的近地表风速、风向差异不显著,其与观测的年平均、季节平均绝对风速偏差在沿岸和下降风区<1 m·s-1,在内陆高原约2—4 m·s-1,年平均风向绝对偏差<10°。

The reliability of the European Centre for Medium-range Weather Forecasts（ ECMWF） ERA-Interim reanalysis and the National Centers for Environmental Prediction（ NCEP） FNL analysis over East Antarctica were investigated by comparing observations of surface pressure,temperature,specific humidity,and winds collected in 2008 along the transverse route from Zhongshan Station to Dome A. Results showed that the surface temperatures of the ERA-Interim reanalysis were closer to observational data than those of the FNL analysis,with a monthly mean absolute deviation 1℃ in Antarctic coastal areas and 2℃ in interior regions. The temperatures of the FNL analysis were significantly warmer than observations on the interior plateau,especially in winter the positive biases can up to8—10℃. Therefore,the FNL analysis can not be directly used to study surface temperature change on the Antarctic Plateau. The surface pressures of the FNL analysis were much closer to observations than those of the ERA-Interim reanalysis,with monthly mean biases of ~ 1 h Pa,while the ERA-Interim reanalysis showed a significant systemic low in coastal regions. The annual / seasonal averaged absolute biases of near surface wind speed between the ERA-Interim reanalysis,the FNL analysis,and observations were less than 1 m·s- 1over coastal and katabatic regions,and about 2—4 m·s- 1over the interior plateau,with absolute wind direction biases of 10°. In addition,the ERA-Interim reanalysis described katabatic winds more accurately than the FNL analysis.