Determination and ranking of trajectory accuracy factors

Trajectory accuracy improvements provide the opportunity to reduce fuel consumption and emissions by increasing the predictability of the National Airspace System (NAS). In addition to the environmental benefits, the ability to improve trajectory prediction accuracy enables trajectory based operations (TBO). Because of the foundational reliance on accurate gate-to-gate four dimensional trajectory (4DT) predictions in TBO, trajectory predictors (TP) will have to meet stringent accuracy performance requirements. There have been significant advances in understanding the accuracy performance and limitations of different TP algorithms. However, implementation of TBO requires identification of the specific aspects of trajectory prediction that will need improvement in order to deliver the necessary accuracy. The challenge is not to select a single trajectory model approach but rather to define which modeling approaches can deliver better performance under different circumstances and to understand the limitations of each. A clear understanding of accuracy factors and the way they affect different modeling approaches allows development of hybrid algorithms that combine the strengths of different approaches. It is known, for instance, that kinetic models are affected by the uncertainty in aircraft mass, while parametric models are affected by the variance in the velocities encoded in lookup tables, but what is the relative contribution of these errors to the overall accuracy of predictions? The paper presents an accuracy factors analysis methodology to rank the sources of error according to their respective impact. This analysis facilitates the identification of the modeling issues that have the largest impact in prediction accuracy and the systematic evaluation of the potential improvements that could be expected from the use of aircraft intent information that will be available when air-ground data link is deployed. The sources of errors in trajectory prediction have been amply studied; however, a clear understanding of the relative contribution of these errors to accuracy performance using realistic scenarios is necessary in order to be able to improve the models. The accuracy factors ranking methodology presented here relies on the computation of the Past Maximum Deviation (PMD) metric for each error measurement and using the PMD to attribute the error to one of four possible error source categories: lateral, vertical, velocity and heading. The PMD metric keeps track of the largest track-trajectory deviation prior to the measurement along each of the four categories. PMD distributions serve as a diagnostic that reveal the performance of a trajectory predictor (TP) along the four dimensions discussed and can be used as an effective tool to compare two TPs or two variants of the same TP. Results of accuracy factor analysis based on the En Route Automation Modernization ERAM parametric algorithm using a live recorded scenario are presented.