The performance of multiple datasets in characterizing the changes of extreme air temperature over China during 1979 to 2012

Using multiple datasets, including three grid datasets (ERA-Interim, WFDEI, and SURF) and one station dataset, we analyzed the climatological distributions and trends of daily maximum air temperature (DMAT) and number of heat days (NHD) over China from 1979 to 2012. In general, all the three grid datasets show the highest climatological DMAT and NHD values in Xinjiang and southeast China, and increasing trends of DMAT and NHD in most of China. Nevertheless, both the climatological condition and trend exhibit differences in detail among the three grid datasets. Using the observed grid dataset (SURF) as a reference, the results show that the WFDEI dataset is closer to SURF than ERA-Interim in characterizing climatological features of DMAT and NHD, and the values from ERA-Interim dataset are lower than the SURF in most part of China. Since the trends of NHD in WFDEI are smaller than those in the other two grid datasets, especially in the middle and lower beaches of the Yangtze River, ERA-Interim is better than WFDEI in representing the trends of NHD. In four regions: Chuanyu, Huanghuai, Xinjiang, and Southeast where the NHD occurs frequently and increases rapidly, the WFDEI is more reliable than ERA-Interim due to the fact that the changes of regional NHD in WFDEI are more consistent with those in observed datasets. Because there exist biases in the records of DMAT between the ERA-interim and WFDEI, it is better to adopt percentile index than absolute threshold to study the changes of China extreme air temperature when using reanalysis datasets.