ERA5 Precipitation over China: Better Relative Hourly and Daily Distribution than Absolute Values

Abstract The temporal distribution of precipitation is of great importance in water availability and the hydrological cycle. Existing studies, directly comparing observed and simulated precipitation amount and intensity over different temporal intervals, show that reanalyses overestimate precipitation frequency and underestimate precipitation intensity, and therefore cannot characterize extreme precipitations. In this study, relative values of precipitation (i.e., the cumulative fractions of precipitation on different percentiles of wettest hours or days to the annual total) were used to evaluate ERA5 over China during the warm seasons of from 1979 to 2015. We found that ERA5 well reproduced the relative values of daily precipitation to the annual total, although hourly results were less satisfied. The cumulative fractions at the hourly time scale at the 99th, 95th, and 90th percentiles were 15.2% ± 3.4%, 38.7% ± 6.6%, and 54.5% ± 7.4% for the gauge observations, and 9.0% ± 1.7%, 27.2% ± 4.0%, and 41.5% ± 5.0% for ERA5, respectively. ERA5 had excellent agreement with gauge observations at daily time scales: gauge observations were 13.0% ± 3.4%, 32.1% ± 5.1%, and 47.6% ± 5.7%, compared to 11.5% ± 3.7%, 30.7% ± 7.0%, and 46.0% ± 8.8% for ERA5, respectively. This difference was mainly due to the high frequency of hourly precipitation in the reanalyses, which were cancelled out at the daily scale. Gauge observations at the hourly scale showed an increase cumulative fraction at the 99th, 95th, and 90th percentiles (0.07% ± 0.03% decade −1 , 0.15% ± 0.03% decade −1 , and 0.17% ± 0.03% decade −1 , respectively) from 1979 to 2015, but did not show an obvious trend at the daily scale. These tendencies were underestimated at the hourly scale but overestimated at the daily scale in ERA5. The results helped us to characterize the temporal distributions and variations of precipitation, and to deepen our understanding of hydrological processes and reanalyses performance.