Homogenization of Monthly Ground Surface Temperature in China during 1961–2016 and Performances of GLDAS Reanalysis Products

Abstract

Maps of observed ground surface temperature (GST) in China generally contain inhomogeneities due to relocation of the observation site, changes in observation method, transition to automatic instruments, and so on. By using the observations of collocated manual and automatic weather stations in China, bias in daily GST caused by the transition to automatic observation systems is corrected for the first time in the present work. Then, the inhomogeneities caused by nonclimatic factors (e.g., relocation of the station and change of observation time) in the historical records of monthly GST are further reduced by using the penalized maximal F-test method. Analysis based on this new homogenized dataset reveals that the trend of annual-mean GST in China is approximately 0.273°C decade?1 during 1961?2016. The warming trend is stronger in winter (0.321°C decade?1) and spring (0.312°C decade?1) and weakest in summer (0.173°C decade?1). Spatially, all the stations in China, except for a few stations in southern China, present warming trends in the annual mean and in spring, fall, and winter seasons. In summer, cooling trends are observed in central and southern China. Moreover, we assess the monthly GST from five reanalysis products of the Global Land Data Assimilation System (GLDAS) during 1980?2016. The warming trends of Noah and the Catchment Land Surface Model (CLSM) from GLDAS-V2.0 are the closest to those of the homogenized observation, while the linear trends in the other three products (Noah, CLM, and MOS) from GLDAS-V1 are obviously different from those of the homogenized observation. Also, it is found that the spatial distribution of the warming trend is substantially overestimated in central China but underestimated in the other regions of China in these five GLDAS reanalysis products.