Wake vortex lidar measurement processing with large-eddy simulations and machine learning

Abstract

Aircraft generated wake vortices may pose a hazard to following aircraft during final approach. Light detection and ranging (lidar) instruments and appropriate processing algorithms are employed for characterizing the position and strength of wake vortices. Unavailable ground-truths in field measurements limit obtainable accuracies of processing algorithms. In this paper the employment of lidar simulations with known ground-truths from large-eddy simulations enhanced by a machine learning based lidar processing algorithm is proposed. This processing algorithm generalization offers wake vortex detection rates above 90% and practical characterization accuracies for comparable simulation and field measurement datasets, decreasing the need for the latter.