Abstract: The upper atmosphere presents a distinct challenge for data assimilation. It is a rapidly changing environment with complex statistical properties. Nonetheless there are a wide range of data assimilation models in development and use across the globe. The fundamental idea of a data assimilation scheme is to make a prediction of the current state of the atmosphere and then update this prediction with data.

However the region is sparsely sampled by data, is strongly driven by solar inputs, and densities can vary by orders of magnitudes from day to night and in response to geomagnetic storms. Since there are a lack of satisfactory covariance models for the ionosphere-thermosphere this provides the impetus to examine the use of ensemble methods of data assimilation. These ensemble approaches, for example the ensemble Kalman filter, also provide the opportunity for producing probabilistic forecasts.

This talk will describe the various approaches to mathematically modelling the ionosphere-thermosphere system, how the output can be used for probabilistic forecasts, techniques to harvest emerging data sources and how this can all be combined to provide actionable space weather products and services into Government and industry.