Calibration of an optimised approach spacing tool


Airports are key members of the air traffic management (ATM) community and network, with local airport ATC performance directly contributing to the overall network performance and delivering benefits to all stakeholders. One of the challenges facing airports and air navigation service providers (ANSP) is to prepare for traffic recovery and future growth in a sustainable way, making efficient use of the runway access.

To support them, at EUROCONTROL we are developing the Calibration of an optimised approach spacing tool (COAST) - an artificial intelligence-based solution to optimise the use of separation delivery tools. This tool will help air traffic controllers (ATCOs) to safely and efficiently manage separation and spacing between arriving aircraft pairs on final approach till the runway threshold, by predicting also the compression effect. It is a time-based solution that can be implemented on traditional distance-based separation (DBS) but achieves the best results applying the time-based separation (TBS) concept under any wind condition.


COAST consists of a machine learning (ML) pipeline which, embedded in the ATC system, provides optimised models and logic to calculate the separation delivery tool indicators to be displayed to the ATCOs. This ML optimised models are trained to account for the effect of variability & uncertainty of aircraft and of wind behaviour on final approach while maximising the delivery arrival throughput gains, even in low wind conditions.

The indicator calculation logic is derived from three ML-based predictive models (a trajectory predictor and one buffer model per indicator) trained at EUROCONTROL using representative largescale historical data of airport arrival traffic and meteorological conditions. A set of conservative models, to cope with aircraft or airlines not frequent enough, are also part of the COAST solution. This guarantees that the safety separation criteria are always satisfied for all incoming aircraft pairs throughout the final approach phase.

More technical details on COAST can be found in our COAST white paper available for download below.

Building on the COAST development experience, EUROCONTROL closely collaborates with EASA in the current development of Guidance Materials on AI assurance for Levels 1 and 2 of Automation. COAST is also taken as a Use case for the EUROCAE Working Group 114 for the development of AI standards in aviation.


The ML pipeline must be built on request and specifically for each airport, under a service agreement (with maintenance fees) with EUROCONTROL. The model training requires historical data of at least one year of final approach traffic data and meteorological information. If you are interested in this service, please contact our experts.