Tests of an Ensemble Kalman Filter for Mesoscale and Regional-Scale Data Assimilation. Part I: Perfect Model Experiments

• DOI: 10.1175/mwr3101.1
• Publication Date: 2006-03-06T16:24:42Z
• Authors: Fuqing Zhang; Zhiyong Meng; Altug Aksoy
• Keywords: MM5; Cyclogenesis

Abstract Through observing system simulation experiments, this two-part study exploits the potential of using ensemble Kalman filter (EnKF) for mesoscale and regional-scale data assimilation. Part I focuses on performance EnKF under perfect model assumption in which truth is produced with same initial uncertainties as those ensemble, while II explores impacts error initiation performance. In first part, implemented a nonhydrostatic [the fifth-generation Pennsylvania State University–NCAR Mesoscale Model (MM5)] to assimilate simulated sounding surface observations derived from simulations “surprise” snowstorm January 2000. This an explosive East Coast cyclogenesis event strong growth at all scales result interactions between convective-, meso-, subsynoptic-scale dynamics. It found that very effective keeping analysis close assumption. The most reducing larger-scale errors but less smaller, marginally resolvable scales. control experiment, was 24-h continuous assimilation typical temporal spatial resolutions reduce by much 80% (compared forecast without assimilation) both observed unobserved variables including zonal meridional winds, temperature, pressure. However, it be relatively efficient correcting vertical velocity moisture fields, have stronger smaller-scale components. domain-averaged root-mean-square after ∼1.0–1.5 m s−1 winds ∼1.0 K comparable or than observational errors. Various sensitivity experiments demonstrated quite successful realistic scenarios tested. will presented II, may significantly degraded if imperfect used, likely case when real are assimilated.