Wind accounts for considerable uncertainty in air traffic, and it often happens in this context that the best predictions available still contain a very significant error. To have a reliable air traffic simulation model, this uncertainty has to be represented, which requires a significant amount of computation. This study presents a way to improve the computing time and memory necessary for Monte Carlo simulations in this application area. The wind-prediction error field is modeled as a multivariate Gaussian stochastic process, with variables correlated both in space and in time. The original algorithm for generating this process, taken from the literature, presented some performance drawbacks; a new algorithm was thus developed using the idea of periodically discarding part of the history and using conditional distributions. This new algorithm performs with remarkable speed and memory gains while being capable of producing results with better temporal autocorrelation than the original algorithm.
Improving Computation of Simulated Wind-Prediction Error for Air Traffic Applications
Journal of Aerospace Information Systems ; 11 , 7 ; 423-432
2014-07-02
10 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Transportation Research Record | 2019
|Traffic flow prediction method based on feature reconstruction error
Europäisches Patentamt | 2020
|Side constrained traffic equilibrium models -- Analysis, computation and applications
Online Contents | 1999
|Europäisches Patentamt | 2021
|