A long-range forecasting model for the thermosphere based on the intelligent optimized particle filtering

The uncertainties associated with the variations in the thermosphere are responsible for the inaccurate prediction of the orbit decay of low Earth orbiting space objects due to the drag force. Accurate forecasting of the thermosphere is urgently required to avoid satellite collisions, which is a potential threat to the rapid growth of spacecraft applications. However, owing to the imperfections in the physics-based forecast model, the long-range forecast of the thermosphere is still primitive even if the accurate prediction of the external forcing is achieved. In this study, we constructed a novel methodology to forecast the thermosphere for tens of days by specifying the uncertain parameters in a physics-based model using an intelligent optimized particle filtering algorithm. A comparison of the results suggested that this method has the capability of providing a more reliable forecast with more than 30-days leading time for the thermospheric mass density than the existing ones under both weak and severe disturbed conditions, if solar and geomagnetic forcing is known. Moreover, the accurate estimation of the state of thermosphere based on this technique would further contribute to the understanding of the temporal and spatial evolution of the upper atmosphere.