Improvement Of Radar Data Assimilation For Radical Velocity In GRAPES-3DVar

Since Doppler radar was applied to meteorology observations and research, the hightemporal and spatial resolutions Doppler radar data had a strong appeal to meteorologist.Earlier applications for Doppler radar radical wind data mainly focused on retrieving windfield base on assumptions of space and time continuity hypothesis. With the developments ofnumerical weather prediction model, assimilation and application of radar radial velocitydevelop fast synchronously. Nowadays, data assimilations of Doppler radar radical windvelocity can be classified as two methods:1）assimilating radar radial wind directly,2）assimilating retrieval wind from radar radical wind. In this paper, we propose a new methodof assimilating single Doppler radar radical wind basing on above two methods. We willintroduce the retrieval wind equations as observation operator, use spatial variation of radicalwind as analytical variables and can change the analysis unit size and shape according tomodel resolution and precision acquirements for resolving the problem of data thinning.Compared with direct assimilation method, the new method introduces the radical windinformation and spatial variation information. As to retrieval assimilation method, theimproved method can avoid using retrieval wind from strong constraint as “observer wind”,and reduce the impact of retrieval error effectively. Basing on the improved assimilationtheory, we develop a module that can assimilate radar radical wind and spatial variationinformation attached to GRAPES3DVAR system. Through the ideal experiment and OSSE(Observer System Simulations experiment), we obtained the following conclusions:1The module attached to GRAPES3DVAR for radar radial wind assimilation withimproved method was constructed and tested, the assimilation system now can assimilate theradial velocity and its spatial varied information.2The improved methods can change analytical unit size according to the accuracy andresolution requirements of model with strong flexibility. It can make full use of highresolution t radar data and solve the problem of data thinning effectively.3The improved assimilation method is based on the direct assimilation and retrievalassimilation, actually, the improved assimilation method is compatible for the directassimilation method and retrieval assimilation method.4The ideal uniform wind field experiment indicate that: assimilation analysis unit is moresmaller, The improved assimilation methods is more similar to direct assimilation, theuniform wind assumption is satisfied more easily, the assimilation of accuracy is also morehigher, but the spatial tangential wind information is less, and if the more bigger the unit ofanalysis is, the more spatial information radial velocity is included, but the impact of non-uniform wind will be more great.5through the OSSE, improving assimilation method for radial velocity and its spatialchanges of information is tested, and the root mean square error of U,V wind field of initialfield is reduced, its effects are better than the direct assimilation and retrieval assimilation. Atlast, we will prove that improved method has huge space on extent and improvement.