[M-GI29] Data assimilation: A fundamental approach in geosciences

[MGI29-02] Enhancement in Assimilation of Doppler Radar Radial Winds

★Invited Papers

*藤田 匡¹、瀬古 弘¹、川畑 拓矢¹、澤田 謙¹、堀田 大介¹、幾田 泰酵¹ (1.気象庁気象研究所)

キーワード：データ同化、観測誤差相関、四次元変分法

Radial wind from Doppler radar is one of the important sources of observational information on detailed atmospheric conditions, often related to severe weather events. This study is aimed to effectively utilize radial wind observations densely distributed in time and space in initializing a numerical weather forecast model. It is important to appropriately handle correlation of the observation error in data assimilation of these high-resolution data. Spatial and temporal correlations of the observation error are statistically diagnosed using the method by Desroziers et al. (2005). The statistical samples are generated running a 3-hourly 4D-Var data assimilation cycle using the experimental system based on the Meso-scale Analysis operated at Japan Meteorological Agency as of 2018 applying the JNoVA 4D-Var (Honda et al. 2005, JMA 2019). The correlation range is found to increase with the beam range and the forecast time, suggesting contributions from errors in the observation operator and the forecast model. An experiment using a simple variational assimilation shows that appropriate handling of the observation error correlation helps to consistently incorporate detailed structures of innovation pattern into the analysis. The correlated observation error of radial wind is applied in a real case experiment based on the Meso-scale Analysis. Spatial and temporal correlation of the observation error is incorporated into the 4D-Var to consistently assimilate radial wind data without applying a severe thinning. In order to extract more information from the dense observation data, the flow-dependent background error is also generated using the Ensemble of Data Assimilation method (EDA; Isaksen et al. 2010), and is introduced into the assimilation system by extending the 4D-Var into the hybrid 4D-Var. Verification of the forecast in the case study shows the flow-dependent background error contributes to effectively utilize the high-resolution and high-frequency radial wind observations. Investigations are also carried out on sensitivity of the performance to the EDA ensemble configuration, influence from the sampling of perturbations added to observations in EDA, and effect from an enhancement in configuration of the ensemble control variables.



Acknowledgements
This work was supported by JST AIP Grant Number JPMJCR19U2, JSPS KAKENHI Grant Number JP 19K23467, and “Program for Promoting Researches on the Supercomputer Fugaku” (Large Ensemble Atmospheric and Environmental Prediction for Disaster Prevention and Mitigation, ID:hp200128/hp120279) of MEXT. This work is based on the operational NWP system developed by Numerical Prediction Division, Japan Meteorological Agency.


References

G. Desroziers, et al., 2005: Quart. J. Roy. Meteor. Soc., 131, 3385–3396.

Y. Honda, et al., 2005: Quart. J. Roy. Meteor. Soc., 131, 3465-3475.

L. Isaksen, et al., 2010: ECMWF Tech. Memo., No. 636.

JMA, 2019: Outline of the operational numerical weather prediction at the JMA. JMA, Tokyo, Japan.