干旱区生态环境知识资源中心

The surface air temperatures measured at 68 meteorological stations in the arid northwestern China during 1979-2012 are compared with temperatures interpolated from the National Centers for Environmental Prediction/Department of Energy (NCEP/DOE) Reanalysis 2 (NCEP R2) and the European Center for Medium-Range Weather Forecasts ERA-Interim. The altitude effects on reanalysis temperature errors are discussed, and the interpolated reanalysis data are calibrated by altitude errors between reanalysis and observation. Using a simple correction method with a constant lapse rate, the elevation-related errors can be greatly removed and an improvement is achieved for the interpolated temperature from both NCEP R2 and ERA-Interim. The cold bias of reanalysis data becomes weak after calibration. On an annual basis, root mean square error of temperature derived from NCEP R2 for each stations has decreased from 6.0 (raw data) to 2.6 A degrees C (calibrated data) and that from ERA-Interim has decreased from 3.2 to 1.4 A degrees C. Similarly, correlation coefficients between raw reanalysis-based and observed temperature are 0.191 and 0.709 for NCEP R2 and ERA-Interim, respectively, whereas the correlation coefficients using the calibrated annual data are 0.819 and 0.932 for NCEP R2 and ERA-Interim, respectively. Generally, ERA-Interim is closer to the ground-based observations than NCEP R2. The topographic correction is more effective in summer than in winter, which may be related to the temperature inversion in winter. Evaluation and correction of reanalysis datasets is a crucial work before the gridded data are applied in climate research, and the altitude-related errors should be calibrated especially in the regions with complex topography.