The proxy of the true atmosphere used for OSSEs is called a Nature Run (NR). Although there are many types of simulation experiments, in a full OSSE the NR is produced by a higher resolution free forecast run that differs from the forecast model used for the data assimilation system (DAS) . It has been realized that the preparation of a NR, including its evaluation, simulation of observations, and it's distribution, consumes a significant amount of effort. It is advantageous that a common NR be used by the various DASs at many institutes: the simulated data are shared. An internationally collaborative effort for full OSSEs has been formed over the last three years.
The first thirteen month long Joint OSSE NRs have been produced by the European Center for Medium-Range Weather Forecasts (ECMWF). The complete data set are saved in NASA/GSFC, NOAA/NCEP, and NOAA/ESRL where they are made available to the wider research community from NASA/NCCS portal system. A part of NR are saved at NCAR as a part of the CISL Research Data Archive data.
Evaluation and preparation for GOES-R and NPOESS will be conducted within the Joint OSSEs. First, the data impact of existing types of data in real and simulated experiments are compared (this process is called the OSSE calibration) and the results are used for interpretation of simulation experiments. All major existing observational data must be simulated for the OSSE calibration. Then new observational data will be simulated and evaluated. The data impact also depends on various components of the DAS and all these components have to be tested and evaluated through full OSSEs. Joint OSSE Nature run showed excellent structure in tropics with realistic hurricanes. Through Joint OSSEs, DAS will be prepared for high resolution data set from GOES-R and NPOESS.
Simulation and assimilation of cloudy radiances will be evaluated. Although both the CRTM and RTTOV can simulate cloudy radiances, cloudy radiances have not been used in DAS. Further development of the CRTM to include cloudy radiances in DAS will be conducted. Modeling subgrid-scale clouds is important for simulating cloudy radiances.
Calibration of the radiance data includes a sampling algorithm which produces a similar distribution of observations as the real data. The adjoint technique is especially useful in the calibration of radiance data, as it allows the skill for an individual channel to be assessed. The skill has to be evaluated for various conditions, as real errors are likely the function of geography, local flow, season, and viewing angle. These errors are also likely to be correlated. The bias, variance, error correlation, and distribution function for the errors have to be modeled in order to be used by any DAS system.