Applicability evaluation of two sets of new high-resolution reanalysis precipitation data on annual and seasonal scale in Hunan

Based on actual measurement precipitation data in Hunan Province, we have made a comparative evaluation of means, interannual variations and correlations for reanalysis precipitation data from China meteorological forcing dataset (CMFD) and Global land surface reanalysis of China for 40 years (CRA-40) in the corresponding period. The results show that (1) the two sets of reanalysis data have better ability to characterize low-rainfall areas in terms of the annual average precipitation distribution. CMFD shows better precipitation distribution in the northwestern Hunan and southern Hunan than CRA-40, while CRA-40 has a better performance in the eastern Hunan than CMFD data. (2) The absolute deviation of precipitation values of CMFD data and CRA-40 data are mainly concentrated in the range of 0-50 mm and 50.1-100 mm. CMFD data in the two ranges accounts for 56.8% and 20%, respectively, and CRA-40 data in the two ranges accounts for 40% and 35.8%, respectively. (3) In terms of the distribution of averaged precipitation in various seasons for many years, most of the averaged precipitation from CMFD are greater than the observations in all seasons, while most of that from CRA-40 are negative deviation in all seasons in northern Hunan and positive deviation in southern Hunan. (4) In terms of the inter-annual variation of the annual mean precipitation, their variation trends for the two sets of data are consistent with the observations in all periods. The annual mean precipitation from 1979 to 1984 and 1999 to 2010 show a decreasing trend, and there is an increasing trend from 1985 to 1998 and 2011 to 2018. Comparing the climate tendency rate of different periods, we can see that the precipitation from CMFD is closer to the observations than that from CRA-40. (5) The correlation coefficients between precipitation from the two sets of reanalysis data and observations are mostly above 0.75, and correlation coefficients at more than 90% of the stations pass the 0.05 significance level test. And the correlation of inter-annual variation rate between precipitation from the reanalysis dataset and the observations is the best in the northern and southern Hunan, and the worse in central Hunan.