The enhancement of atmospheric drag prediction using space-tracking data for accurate debris surveillance and collision warning

摘要

This research investigates the temporal variation of the coefficient of atmospheric drag value (CD) and its subsequent effects on orbit prediction. Atmospheric drag is one of the most dominant forces exerted to space objects at altitudes below approximately 1500 km in the Low Earth Orbit (LEO). Its accuracy is governed primarily by the accuracy of the atmospheric mass density modelling, area-to-mass ratio and the CD value. Traditionally, the tracking of space objects is conducted using the radar method and the optical approach. Recently, Satellite Laser Ranging (SLR) technique is being developed to track space debris. However, it is still limited by the most fundamental problem of unable to accurately predict the motion of space objects, which is largely due to the insufficient accuracy of determining the atmospheric drag. The focus of this research is to investigate viable approaches to enhance the prediction of the CD value for higher accuracy prediction of orbits of space objects. The conventional CD value prediction approaches, i.e., the fix 2.2 CD and variable CD methods, are investigated. The more accurate variable CD approach has presented a repetitive cyclical change in the estimated CD values over the study period from 2004 to 2006 using Stella as the experimental satellite. This suggests a different scenario to the fixed value of 2.2 approach commonly adopted by the space industry. Due to the repetitive cycle of the CD variations, Fourier series are selected to fit the estimated CD values over the study period. The fitting function is extrapolated to predict CD values for 2007, which are subsequently applied to the subsequent orbit prediction process using the fix CD value method. This implies that the predicted CD values are pre-determined prior to the orbit determination and prediction, similar to the fix 2.2 method that adopts a fixed value of 2.2. The orbit prediction results using the fitting function have demonstrated significant improvements over the traditional fixed 2.2 CD value method. The fitting function approach is also verified by performing the same experiments to satellites Starlette and ERS-2, where noticeable improvements in the orbit predictions are also achieved. For orbit prediction, the fixed value method has shown to be more computationally efficient since approximate 20% reduction in data processing time is achieved compared to the more accurate variable CD approach. This is one of the fundamental reasons for the space industry to adopt the fixed value method, especially when timely prediction of orbits is the primary goal to many orbit applications. This research has presented the fitting function approach for CD value prediction and the results have demonstrated that higher accuracy orbit predictions without degradation to the efficiency are achieved compared to the fix 2.2 CD method. Thus, this research will provide a valuable performance assessment of the conventional and the fitting function CD value estimation/prediction approaches for atmospheric research. In addition, it will also offer constructive guidance to minimise the limitations currently confronted by the space debris tracking, specifically atmospheric drag prediction.