Development and Evaluation of East Asia Regional Reanalysis Based on the Advanced Hybrid Gain Data Assimilation Method Using WRF Model

Abstract:

As the importance of a high-resolution regional reanalysis emerged, a regional reanalysis system is needed to produce regional reanalysis dataset over East Asia. Since hybrid data assimilation (DA) methods were developed and implemented by many organizations and institutes over the world, a hybrid gain DA method has been developed. The hybrid gain DA algorithm is a simple and practical method to take advantage of each strength of two different methods without much effort, weighting analyses from variational- and ensemble-based DA algorithms, not their background error covariances.

In this study, based on the hybrid gain DA method, an advanced hybrid gain algorithm (AdvHG), which combines E3DVAR and ERA5 based on WRF model, is newly proposed. This method, which uses the existing reanalysis dataset of high quality, is efficient because of not only cost savings but also the use of the state-of-the-art reanalysis from ECMWF assimilating all available observations including satellite radiance. Therefore, in this study, East Asia Regional Reanalysis (EARR) system is developed based on the advanced hybrid gain DA method with WRF model. With this system, the high-resolution regional reanalysis and reforecast with 12 km resolution are produced over East Asia for the six-year period of 2013–2018.

The EARR is evaluated with E3DVAR, ERA5, and ERA-I for January and July in 2017, respectively. In general, for upper air variables, ERA5 outperforms EARR, whereas EARR outperforms ERA-I for January and shows comparable performance to ERA-I for July. On the contrary, EARR better represents precipitation than ERA5 (WRF-based) as well as ERA5_fromECMWF (precipitation reforecast of ERA5 from ECMWF) for January and July in 2017. Therefore, though the uncertainties of upper air variables of EARR should be considered when analyzing them, the precipitation reforecast of EARR is more accurate than that of ERA5 for both two seasons.

Acknowledgments

This work was supported by a National Research Foundation of Korea (NRF) grant funded by the South Korean government (Ministry of Science and ICT) (Grant 2017R1E1A1A03070968) and the Korea Polar Research Institute (KOPRI, PN20081). The authors gratefully acknowledge the late Dr. Fuqing Zhang for providing the resources and discussions at the earlier stages of this study.