The Global Modeling and Assimilation Office at NASA has been participating in the international Observing System Simulation Experiment organized by ECMWF and NCEP. For our contribution to this effort, we have produced and validated a baseline set of simulated observations for summer and winter periods. These are drawn from the 13-month nature run provided by ECMWF at T512L91 resolution. Their production has been expedited by introducing several simplifications appropriate for present applications; e.g., regarding the effects of clouds on infrared radiance observations.

Validation has been conducted using the NCEP/GMAO 3DVAR (GSI) data assimilation system. This includes tuning of some characteristics of the simulated observations and their instrument plus representativeness errors. The procedure exploits the GSI adjoint to efficiently estimate the reductions in 1-day forecast error due to individual sets of observations. Our goal has therefore been to match these estimates in the OSSE context with those from a corresponding assimilation of real observations. The algorithms and results of our simulation technique and validation procedure will be described in this presentation.