2004 , Volume 9, Special issue, p.34-41

An Atmospheric Data Assimilation System (DAS) is described. DAS is based on the semi-Lagrangian variable resolution atmospheric model (developed in the Institute of Computational Mathematics RAS) and the optimum interpolation objective analysis based on the incremental approach (developed in the HydroMetCenter of Russia). The methods and technology of the implementation of data assimilation and the ensuing forecasts (forecast lead time up to 5 days) on the basis of the DAS analysis are presented. Computations are distributed between a mainframe CRAY-YMP and a cluster of 4 servers, where each server contains 4 Intel Itanium II processors. DAS is used not only for assimilation of operational observations, but also is utilized as a tool for research activity. In particular, the investigation of the influence of observational errors on the data assimilation is briefly described.