A three-dimensional variational analysis (3DVAR) system applied in physical space, gridpoint statistical interpolation (GSI), is currently being developed at the National Centers for Environmental Prediction (NCEP) in collaboration with the Global Modeling and Assimilation Office at NASA Goddard. The GSI is a unified analysis system that allows for both global and regional applications, and can currently be used to generate analyses for a variety of forecast models and domains, including the NCEP Global Forecast System (GFS), the Weather Research and Forecasting (WRF) model, and others. The fact that the GSI is formulated in grid space allows for more degrees of freedom in defining background error statistics than a system formulated in spectral space.

This presentation will show results from various experiments utilizing the global version of GSI, by comparing with results from the global analsysis system which is currently operational at NCEP, the spectral statistical interpolation system. The importance and sensitivity to the specification of background error statistics will be shown.