Can temperature extremes in China be calculated from reanalysis?

Based on daily maximum, minimum and mean surface air temperature from National Centers for Environmental Prediction/National Center for Atmospheric Research Reanalysis (NCEP/NCAR) and European Centre for Medium-Range Weather Forecasts (ECMWF) reanalyses, the distributions of twenty temperature indices are examined in China during 1958-2011. ECMWF includes ERA-40 for the period 1958-2001 and ERA-Interim during 2002-2011. The consistency and discrepancy of extreme indices between reanalyses and observations (303 stations) are assessed. In most cases, temperature indices between NCEP/NCAR and ECMWF have good agreements. For both reanalysis, cold days/nights have decreased, while warm days/nights have increased since 1980. Temperatures of the coldest days/nights and warmest days/nights significantly increase over the entire China, and the diurnal temperature range demonstrates slight variations; the amounts of growing season length, and summer/tropical days have increased, consistent with the decrease in numbers of frost/ice days. Furthermore, the persistence of heat wave duration and warm spell days has increased and consecutive frost days have reduced. Meanwhile, consecutive frost days, cold wave duration and cold spell days from NCEP/NCAR have decreased and consecutive frost days have increased, while these indices from ECMWF turn to the opposite directions. Compared with observations, temperature extremes from two reanalyses have small relative bias and the root mean squared errors, while correlation coefficients are positively high. These suggest that both reanalyses can reproduce the variability of temperature extremes obtained from observations, and can be applied to investigate climate extremes to some extent, although the biases exist due to the assimilation differences.