期刊名称：Tellus A: Dynamic Meteorology and Oceanography
电子版ISSN：1600-0870
出版年度：2015
卷号：67
页码：1-17
DOI：10.3402/tellusa.v67.26495
语种：English
摘要：Limited-area models (LAMs) allow high-resolution forecasts to be made for geographic regions of interest when resources are limited. Typically, boundary conditions for these models are provided through one-way boundary coupling from a coarser resolution global model. Here, data assimilation is considered in a situation in which a global model supplies boundary conditions to multiple LAMs. The data assimilation method presented combines information from all of the models to construct a single ‘composite state’, on which data assimilation is subsequently performed. The analysis composite state is then used to form the initial conditions of the global model and all of the LAMs for the next forecast cycle. The method is tested by using numerical experiments with simple, chaotic models. The results of the experiments show that there is a clear forecast benefit to allowing LAM states to influence one another during the analysis. In addition, adding LAM information at analysis time has a strong positive impact on global model forecast performance, even at points not covered by the LAMs.
关键词：Ensemble Kalman Filter; limited-area models; composite state
