airsecoma GmbH analysed Gigabytes of air traffic control data to reveal which metrics can be used to take better decisions in a safety-critical operations room.

In Air Traffic Management (ATM), control room supervisors (SV) monitors the upcoming air traffic load using the EUROCONTROL NM CHMI software and/or other similar tools.

The main “metrics” to evaluate the traffic load and is commonly expressed as Entry Counts (EC) over a period (most often Hourly Entry Counts - HEC) - or Occupancy Counts (OCC). These metrics all represents a different way to count aircraft within a sector or traffic volume (TV).

This last metric, i.e. OCC, is today of particular interest to the supervisors or Flow Manager Positions (FMPs), as assumed to better reflect the ATCO workload as EC or HEC and be more precise with a time step every minute.

Nevertheless, when using Occupancy, a “duration” variable shall be defined by the user. This duration, e.g. 1, 10 or 15 minutes, corresponds to the time window during which aircraft entering or being a sector are counted.

This variable has a significant impact on the counts calculated over time and how decisions are made to split or merge TVs by the supervisors or Flow Manager Positions (FMPs).
Therefore, this variable has a critical importance for the implementation of concepts such as Dynamic Airspace Sectorisation

It has been found that no consensus exists amongst Air Traffic Control Centers and their supervisors on which duration is the best in operations. Nevertheless, most differences among the supervisors seem rather based on intuition and personal feeling than on hard facts.

For this purpose, airsecoma was tasked to determine based on a quantitative analysis which duration is the most suitable to take decisions in operations.

Common claims to validate

Longer Occupancy Durations:
  • “gives better predictions earlier than shorter occupancy duration (better here being more accurate in relation to final occupancy counts)”
  • “are less volatile predictions than shorter occupancy duration (volatile meaning subject to sudden changes)”

Study objective - How accurate and volatile are air-traffic control predictions?

The main objectives of this present study were to validate or invalidate several assumptions, by comparing the different metrics both qualitatively and quantitatively using the following indicators for each occupancy duration:

  • The traffic predictability, i.e. the quality of the prediction
  • The traffic prediction volatility, i.e. the changes over time of the prediction


Technologies used enabled an easy processing and visualisation of a very large amount of data.

Study methodology

To validate the previously mentioned hypothesis, occupancy counts with different duration (1, 10, 15) were retrieved for a selected sub-set of TVs over a sample period (2 weeks in January and February) and store in Azure using airsecoma EUROCONTROL NM B2B Web Services CONNECTOR

To visualise and analyse the recorded data together with all calculated statistical indicators, an application was developed in Power BI. This application contains several pages or “reports” and enables the users to select different criteria for analysis (e.g. LAT, date, TVs, etc.).

Analysing traffic prediction accuracy consists in measuring the error between the predicted value, and the final counts. However, to be able to compare different traffic volumes and occupancy durations, several statistical processes (e.g. normalisation) and metrics were used to quantify prediction errors.


airsecoma GmbH
Schröderstr. 3
10115 Berlin

Tel: (+49) 17 36 09 65 92

airsecoma GmbH analyses Gigabytes of air traffic control data to reveal which metrics can be used to take better decisions in a safety critical operations room.