资源环境科技发展态势分析平台

The near surface air temperature is the primary indicator for climate change. Reanalysis as the surrogates for large-scale observations are widely used in the Tibetan Plateau because of the sparse meteorological network. However, an average bias of -3.54 degrees C and root-mean-square error (RMSE) of 4.31 degrees C were found between ERA-Interim monthly 2-m temperature and observation over the Tibetan Plateau, which indicated that a correction procedure for ERA-Interim is necessary before local scale applications. To overcome this challenge, a robust elevation correction method is developed to downscale ERA-Interim 2 degrees x 2 degrees monthly 2-m temperature data based on ERA-Interim internal vertical lapse rates. This method is validated against 80 meteorological stations from 1979 to 2013 located in 26 ERA-Interim grid cells. It is also compared with other four correction methods, which are using different lapse rate schemes such as fixed monthly lapse rates, surface lapse rates calculated from the meteorological stations (within a single grid or with neighbouring sites), as well as a third-order curvilinear function of ERA-Interim pressure level data. The results indicate that the correction method using ERA-Interim internal vertical lapse rates cannot only significantly reduce the bias (89%) and RMSE (62%) for the original ERA-Interim data, but also capture the inter-annual variations for the plateau-wide climatology very well. The seasonal and annual temperature warming trends are also modelled encouragingly compared with other four methods. The strongest advantage of this method is that it is independent of local meteorological stations. Therefore, it is possible to extrapolate ERA-Interim temperature data for any other high mountain areas where no measurements exist. This work will help the scientific community identify the most proper and easiest method to downscale reanalysis temperature data for climate impact assessments at the site or regional scale.