While past studies have investigated the effect of neutral atmospheric density mis-modeling on satellite conjunction assessment (CA), none has focused their investigation specifically on serious (high-risk) conjunction events, which are the event types that drive both risk and workload for CA operations. The present study seeks to do this by reprocessing a large number of archived actual conjunction events, artificially introducing atmospheric density error to these events, and then examining the effect of these introduced errors on the probability of collision (Pc) calculation, which is the principal parameter used to assess collision risk. These reprocessed calculations are executed both with the satellites’ covariances unaltered and with a covariance modification that accounts for the induced atmospheric density error. The results indicate that the situation is greatly aided by an a priori knowledge of the approximate density estimation error, even if the model itself is unaltered—missed detections (Type II errors) due to density estimation uncertainty are substantially reduced when the density model prediction error is characterized and can be included in the satellite covariance and thus Pc calculation. Overall improvements in density model predictive performance are also important to improving CA, especially for false alarm (Type I error) reduction; but model enhancements that include a robust, in-model error analysis offer the most significant improvements to the CA enterprise.
PDF
download
878987797
028-85220240
