CRA-40/Atmosphere-The First-Generation Chinese Atmospheric Reanalysis (1979-2018): System Description and Performance Evaluation

Atmospheric reanalysis reproduces the past atmospheric conditions through assimilation of historical meteorological observations with fixed version of a numerical weather prediction (NWP) model and data assimilation (DA) system. It is widely used in weather, climate, and even business-related research and applications. This paper reports the development of CMA's first-generation global atmospheric reanalysis (RA) covering 1979-2018 (CRA-40; CRA refers to CMA-RA). CRA-40 is produced by using the Global Spectral Model (GSM)/Gridpoint Statistical Interpolation (GSI) at a 6-h time interval and a T(L)574 spectral (34-km) resolution with the model top at 0.27 hPa. A large number of reprocessed satellite data and widely collected conventional observations were assimilated during the reanalyzing process, including the reprocessed atmospheric motion vector (AMV) products from FY-2C/D/E/G satellites, dense conventional observations (at about 120 radiosonde and 2400 synoptic stations) over China, as well as MWHS-2 and GNSS-RO observations from FY-3C. The reanalysis fitting to observations is improved over time, especially for surface pressure with root-mean-square error reduced from 1.05 hPa in 1979 to 0.8 hPa, and for upper air temperature from 1.65 K in 1979 to 1.35 K, in 2018. The patterns of global analysis increments for temperature, specific humidity, and zonal wind are consistent with the changes in the observing system. Near surface temperature from the model's 6-h forecast reflects the global warming trend reasonably. The CRA-40 precipitation pattern matches well with those of GPCP and other reanalyses. CRA-40 also successfully captures the QBO and its vertical and temporal development, hemispherical atmospheric circulation change, and moisture transport by the East Asian summer monsoon. CRA is now operationally running in near real time as a climate data assimilation system in CMA.