A New Method of Physics-Based Data Assimilation for the Quiet and Disturbed Thermosphere

Abstract

The ability to accurately track and predict satellite locations is of paramount importance to space-faring nations. In the low Earth orbit satellite environment, atmospheric drag is by far the dominant error associated with orbit propagation. Nowcasts of thermospheric density are routinely accomplished through calibration of semiempirical models using recent data, yet forward predictions degrade quickly as lead time increases. Physics-based approaches offer a great forecasting potential but one that has yet to be realized due to a lack of robust data assimilation schemes. In an effort to account for the driver/response characteristics of the thermosphere-ionosphere system, a new data assimilative technique is developed. Abandoning the ensemble Kalman filter framework in favor of a variational technique, iterative model reinitialization is applied self-consistently to estimate a time history of effective solar and geophysical drivers. The current implementation of this technique, referred to as Iterative Reinitialization, Driver Estimation and Assimilation, works by ingesting neutral mass density measurements from low-Earth orbiting accelerometers. A long-term simulation is carried out during 2003, a period consisting of a wide range of solar and geomagnetic activity levels. The new technique is shown to greatly reduce RMS errors of the physics-based model relative to ingested observations from the Challenging Mini-Satellite Payload (CHAMP) satellite as well as to an independent validation data set from the Gravity Recovery And Climate Experiment (GRACE) satellite. This work is the first such demonstration of an accurate and robust physics-based method capable of specifying neutral density during both quiet and disturbed times and offers a promising outlook for improving density forecasting capabilities.