Much as aircraft are affected by the prevailing winds and weather conditions in which they fly, satellites are affected by the variability in density and motion of the near earth space environment.  Drastic changes in the state of the upper atmosphere, caused by geomagnetic storms or other phenomena, result in perturbations of LEO satellite motions through drag on the satellite surfaces. This can lead to increased position uncertainties, larger covariance ellipses, and errors when predicting collisions in space. As the population of satellites in Earth orbit grows with increased commercial activity, higher space-weather prediction accuracy is required for critical missions, such as accurate catalog maintenance, collision avoidance for manned and unmanned space flight, reentry prediction, satellite lifetime prediction, defining on-board fuel requirements, and satellite attitude dynamics.

We describe the next-generation commercial nowcast and forecast system for specifying the neutral atmospheric state related to orbital drag conditions.  This tool is based on several state-of-the-art upper-atmosphere models running in real-time and uses assimilative techniques to produce a satellite drag nowcast. This software will also produce 72 hour predictions of the global satellite-drag conditions using near real-time and predicted space weather data and indices as the inputs.  Features of this technique include:

• Satellite drag specifications with errors lower than current models
• Altitude coverage up to 1000km
• Assimilation of satellite drag and other datatypes
• Real time capability
• Ability to produce 72-hour forecasts of the atmospheric state

In this paper, we summarize the model design and assimilative architecture, and present preliminary validation results.  Validation results will be presented in the context of satellite orbit errors and compared with several leading atmospheric models.  As part of the analysis, we compare the drag observed by a variety of satellites which were not used as part of the assimilation-dataset.