Support algorithm for air traffic controllers’ arrival spacing: Improvement of trajectory estimation using Gaussian Process Regression

Abstract

A flying aircraft generates wake turbulence behind it from its wingtips. To avoid the effects of wake turbulence, successive arrival aircraft at a runway must maintain appropriate longitudinal separation between the leader and follower aircraft. The wake turbulence separation minima are applied based on an air traffic controller’s instructions. This task, arrival spacing, place a high workload on air traffic controllers. To reduce the workload on air traffic controllers, this study proposes a support algorithm for arrival spacing. The support algorithm provides an indicator for each aircraft. The indicator is calculated to maintain an arrival separation that matches the separation minima; thus, if the air traffic controller adjusts the aircraft’s position to match the indicator, the arrival aircraft can maintain proper separation. The accuracy of the trajectory estimation in the support algorithm is crucial for the support algorithm to provide an accurate indicator. To configure an accurate trajectory estimation, an aircraft’s speed profile is generated by applying Gaussian Process Regression (GPR). Actual operational data obtained from radar data is used as input data for the regression. To evaluate the capabilities of the support algorithm, case studies are conducted using radar data. The results show that the support algorithm can calculate the appropriate indicator near the runway threshold. However, the indicator has errors in the initial phase of arrival spacing. By applying the speed profile generated by GPR to the support algorithm, the average of the indicator error reduces to about one-fifth when applying the speed profile based on an airline procedure model.