Uncertainty Estimation Strategy for Grouped Target Tracking under Sparse Observation

Uncontrolled reentry debris poses significant challenges to trajectory prediction due to its nonlinear dynamics, sparse observations, and environmental variability. This study proposes a hierarchical uncertainty estimation framework that integrates differential propagation, statistical modeling, and multi-source data fusion. By introducing the Atmospheric Centrifugation Effect and classifying debris into Pioneering Detection Layer (PDL), Forefront Propagation Layer (FPL), and Rear Propagation Layer (RPL) based on the area-to-mass ratio (A/M), the model captures stratified propagation patterns. Odd- and even-order nonlinearities are explicitly decomposed to characterize asymmetric drift and uncertainty expansion. Edgeworth and Gram-Charlier A series are employed to model heavy-tailed distributions, while kernel-based convolution enhances sparse target extraction. Simulation results using NASA’s breakup model and environmental data from multiple launch sites demonstrate the effectiveness of the proposed framework in improving trajectory correction and hazard zone prediction. This research provides theoretical and practical support for space debris risk management under limited observation conditions.