研究期間：10108~10207;This proposal seeks to improve the regional mesoscale ensemble Kalman filtering (EnKF) under the condition of strong nonlinearity and interactions between coupled synoptic-meso/small scale dynamics. When using the multi nested-domains for regional data assimilation, different model grid resolutions aim to represent the characteristics of different scales of the atmospheric flow. Correspondingly, errors are characterized by different scales and thus analysis corrections should consider such error characteristics. However, this may require high computation cost for assimilating different type of observation networks, e.g. the dense observations like radar data for high-resolution grid vs. the sparse observation like the station data. A weight interpolation method (Yang et al. 2009) with the WRF-LETKF system is proposed to effectively assimilate the dense observations with high-resolution grids without sacrificing the analysis accuracy of the high-resolution model grid and incorporate these weights into other domains. By constructing the multi-scale weighting maps, we propose to derive analysis for nested domains with better dynamical consistency. In addition to the high efficient LETKF analysis, we propose to establish the ensemble Kalman smoother for nested-domains for applying the outer-loop method (Yang et al. 2010) in order to adjust the nonlinear dynamical evolution and improve the interactions between synoptic and meso/small scales. We will investigate whether such dynamicallyconsistent initial conditions will be beneficial when performing regional prediction for typhoons or other severe weather events.
