5.3.1. Scenarios

The real datasets contain 322, 421, and 233 Malpensa flights. They were manually prepared and then read with RouGe Software V1.11 from DLR (Braunschweig, Germany). All military flights, helicopters, departures, and parachutist transports were filtered. To validate the data visually, the datasets were converted to the Google Earth KML data format. Figure 14 illustrates the flown trajectories on 11 May 2019 that occurred under adverse weather conditions. One can recognize that multiple holdings were flown due to thunderstorms at and around the TMA area. Figure 15 shows the flown trajectories on 6 August 2019 under normal undisturbed weather conditions.

The FlightRadar24 data prepared as simulation scenarios contain the following information about callsign, aircraft type, weight category, position data plus time, speed, flight altitude, and heading.

The constructed simulation scenarios were each planned using an AMAN with no adverse weather as a baseline and weather forecasts from different forecast models for two different severe weather events. Flight paths, times, the number of clearances, and fuel consumption were used as key performance indicators (KPI) to compare these models. The considered traffic scenario is based on the dataset from 11 May 2019. Seven scenarios were planned using an AMAN and simulated:


**Figure 14.** Prepared FlightRadar24 datasets from 11 May 2019 with flights operated under adverse weather conditions.

**Figure 15.** Prepared FlightRadar24 datasets from 6 August 2019 with normal undisturbed weather conditions.

5.3.2. The Malpensa Airspace Implementation for the AMAN

For aircraft arrival planning and simulations, MXP airspace had to be implemented in DLR's AMAN and the traffic simulation environment (Figure 16). For this simulation, the airspace does not need all of the waypoints and sector boundaries of the real airspace; therefore, only waypoints relevant for arrivals have been inserted. Departure routes were not implemented, and this display thus looks less complex and simpler than usual radar images from MXP. For sequencing, all Standard Arrival Routes (STARs) with the related Flight Management System (FMS) waypoints were assigned as specified in the Aeronautical Information Publications (AIPs) of MXP [54]. The waypoints are connected with constraints regarding maximum and minimum altitudes, as well as maximum and minimum speeds. These data are needed for the 4D trajectory calculation since it is not only speed but also flight altitudes that have an influence on flight times until touchdown.

**Figure 16.** The Malpensa airport and airspace in the AMAN planning and simulation environment.

MXP has two runways in the south–north direction: 35L and 35R. Because of the closeness of the Alps, landings are exclusively northbound. With seven Entry Fixes or Metering Fixes, the TMA has an average number of entry points that, while not covering all points of the compass equally, are a good compromise with respect to the Alps, other nearby airports, and restricted military areas in the vicinity of the airport. The Path Stretching Area (PSA) to the south of the airport is formed by a trombone to the east and a double-trombone to the west, which allows for an even distribution of approaches across the two runways. According to the AIP, it is not until the Initial Approach Fix (IAF) with the name INLER on Final that a decision is made as to which runway aircraft will be guided to; however, analysis of radar data has shown that controllers guide aircraft to the appropriate centerline much earlier. Runway switches on centerline or final are rather the exception.

STARs are assigned for each aircraft based on the respective approach direction and, from the north, are also based on utilization of the western and eastern trombones. Aircraft from the north are routed accordingly via the ODINA and RIXUV fixes, aircraft from the east are routed via EVRIP and PEXUG, aircraft from the southeast are also routed via PEXUG, aircraft from the south are routed via MEBUR, aircraft from the southwest are routed via DEVOX, aircraft from the west are routed via ASTIG, and aircraft from the northwest are routed via NELAB. During the selected days for the validation, all routes were active and used by approaches.

#### **6. Discussion of Approach Planning Results under Severe Weather Conditions**

For validation of the planning and display software, the described scenarios were calculated and the results were then presented to an international group of ATCOs. They evaluated the individual scheduling and trajectory results of the AMAN 4D-CARMA, as well as the entire approach of the support system.

#### *6.1. Results of Diversion Calculations and Sequence Planning*

For validation of the new support functions implemented in DLR's AMAN, an evaluation was conducted with the AMAN that was linked to a traffic simulation and a questionnaire. The extended AMAN runs in 15 separate modules that communicate with each other via a Maria database. In addition, the DLR's radar display RadarVision is used, which can display aircraft movements as well as additional information on flight plans and the generated trajectories. In addition to the 2D route, this includes the planned altitudes and speed profiles and, during a simulation, the current air situation in relation to the planned 4D trajectory. For traffic simulation, the AMAN was coupled with DLR's ArrOS system. ArrOS enables precise movements along a trajectory and simulates cooperative controllers and pilots. A selection of screenshots from the simulation runs are used to illustrate the possibilities and results of the validation trials with 4D-CARMA. The following pictures are screenshots showing selected traffic and weather situations with different weather nowcast models.

Diversion route calculation has to consider different constraints in the TMA and the surrounding sectors. In the TMA, the routes are much more convergent, and the aircraft have to overfly more significant waypoints with specific constraints than in the adjacent sectors. As a result, the AMAN can plan larger-scale diversions in the sectors. In Figure 17, the AMAN plans an early diversion and guides the aircraft EJU34CD on a more southern route than the STAR intends.

**Figure 17.** A large-scale diversion route plan for aircraft EJU34CD, currently No. 1 in the arrival sequence and approaching MXP from the west. The white arrows indicate the original STAR and the yellow–red dotted line represents the AMAN 4D trajectory. MXP traffic mix from 11 May 2019 with weather from 11 May 2019 and the WRF-RUC nowcast model.

For traffic visualization on radar screens, a dark image is preferred by some air traffic controllers. In Figure 18, a dark color scheme was selected with trajectories and a thunderstorm cell. In order to reduce the visual presence of the weather information, some controllers had suggested in the run-up to the SINOPTICA concept's development that convective zones should only be displayed as bordered areas [27]. RadarVision now allows controllers to choose between the two weather presentations of filled and bordered areas and also allows for switching between dark and light-colored traffic visualizations through keyboard input.

**Figure 18.** The dark color scheme with convective zones displayed as bordered areas. Malpensa traffic mix from 11 May 2019 with weather from 11 May 2019 and the WRF-RUC nowcast model.

A particular challenge was route planning that required diversions around several convective zones moving at different speeds in different directions. In Figure 19, an example of a diversion around four smaller convective areas is plotted for aircraft BEE156H arriving at Malpensa from the northwest. In weather situations of this type, the AMAN must decide independently whether it is more appropriate to fly through the loosely distributed convective zones or to fly completely around the entire affected area.

**Figure 19.** Route calculation for aircraft BEE156H approaching MXP from the northwest through a loose group of convective areas. If several convective cells occur as a group, the AMAN tries to guide aircraft around them as a whole. MXP traffic mix from 11 May 2019 with weather from 6 August 2019 and the RaNDeVIL nowcast model.

In Figure 20, a 4D trajectory was planned with a diversion right through a convective area. However, following the weather forecast, this cell shifted to the southeast in the course of the following minutes so that, when the aircraft arrived, the route precisely avoided this area. If convective cells appear as a loose group, the AMAN tries to find a straight-line route through them. It can happen that routes approach each other for up to a few miles but are continued differently due to a small spatial and temporal distance (Figure 21).

**Figure 20.** Route calculation for the aircraft EJU36LR approaching MXP from the northeast through a loose group of convective areas. Currently, this route is still blocked by a convective cell; however, based on the forecast, the AMAN knows that the route will be clear by the time the aircraft arrives. MXP traffic mix from 11 May 2019 with weather from 6 August 2019 and the RaNDeVIL nowcast model.

**Figure 21.** Two different route calculations for aircraft approaching Malpensa from the northeast through a loose group of convective areas. Even a small variance in temporally different initial conditions can lead to different optimal routing. Malpensa traffic mix from 11 May 2019 with weather from 6 August 2019 and the RaNDeVIL nowcast model.

All routes calculated and displayed in this paper are 4D trajectories that include altitude, speed, and time in addition to position. They are always calculated on the basis of EUROCONTROL's BADA 3.13 [36] and are therefore realistic within the scope of technical possibilities. They should also therefore be able to be safely flown by the respective types of aircraft. BADA data for the respective aircraft types are also used for sequence calculation on the finals and the runways so that the distances between the aircraft follow the separations corresponding to their weight classes. This is also valid for the 4D trajectories calculated as diversion routes around severe weather areas. As an example, an aircraft approaching from the north and its speed and altitude profile are presented in Figure 22. It can be seen from the altitude profile in the upper diagram that it is not possible for an aircraft to fly over or under a thunderstorm cell due to the descent that is usually already initiated at the distance of a few dozen miles from the airport.

**Figure 22.** Visualization of an AMAN-calculated 4D diversion trajectory for aircraft BEE156H approaching MXP from the west and currently ninth in the arrival sequence. In the upper dark blue shaded diagram, the altitude profile is displayed, and in the lower one the speed profile is displayed. It can be seen that the aircraft is already descending as it flies around the thunderstorm cell. MXP traffic mix from 11 May 2019 with weather from 11 May 2019 and the PhaSt nowcast model.

Technically, 3D polyhedrons could be used for weather areas to specify the upper and lower boundary of convective cells; however, this does not bring any operational advantages in the context of approach planning. These visualizations indicate the strict dependence of AMAN-calculated 4D trajectories under the influence of local weather on the applied weather model and its assessment of meteorological severity for aviation.

#### *6.2. Results of Controllers' Validations*

In the validation process at the end of the project, five active and former ATCOs from Austria, Germany, and Poland received an introduction to the aims and the results of the functional extended AMAN with severe weather guidance support. Afterwards, an AMAN demonstration was held consisting of six videos comprising different AMAN- planned trajectories that changed depending on weather dynamics, traffic dynamics, and the position of the aircraft. Additionally, two different kinds of weather presentations (filled and bordered areas) on radar displays and 4D trajectories with predicted speed and altitude profiles were introduced.

During the validations, the main idea of combining meteorological data, air traffic approach procedures, and the topography of Malpensa in an AMAN for arrival support was presented. The new approach with the adjustment of routes at early stages through the use of forecast data and the organization of air traffic around dynamic weather areas was explained. Five new supporting functionalities with routing, sequencing, weather animation, aircraft tagging in the form of labels, and guidance advisories were introduced and demonstrated. After the introduction, the controllers viewed and analyzed the videos demonstrating AMAN planning and its impact on live traffic.

The controllers then filled out a questionnaire with a total of 17 main questions and sub-questions on the technical assessment. Similar to the user requirements survey, 14 questions offered a seven-point response option that ranged between "strongly agree" and "strongly disagree". Each question also had a comment field where additional comments and suggestions as well as explanations of the answers could be added. The controllers had no time constraints when completing the survey. The questionnaire consisted of 17 questions and sub-questions, 14 of which had graded response options and could be statistically analyzed:

	- a. In the entire TMA?
	- b. In arrival sectors?
	- c. On anything else?
	- a. In the entire TMA?
	- b. In arrival sectors?
	- c. On anything else?
	- a. Would it lead to timelier planning?
	- b. Would it anticipate decision making?

Analysis of the questionnaires showed that the controllers had professional experience ranging between 1 and 25 years, which thus covers a wide range of experience in the profession of air traffic control. It becomes clear that the newly developed support and visualization functions received clear overall approval from the controllers. Not all controllers were able to answer all the questions because their training and work experience did not give them sufficient knowledge of all the work positions where severe weather support could conceivably be provided (by their own appraisal). Particularly important for our controllers' support system evaluation were the comments, suggestions, and criticisms

of the participants. For this reason, a generous comment field was added to each question, which was also used by most participants. These comments were intended to further develop the system so that it could also be operationally used in the future. In addition to a general level of approval, this kind of open user feedback is also important to ensure that further developments do not neglect the needs and requirements of air traffic controllers.

Evaluation of the responses showed the overall comprehensive satisfaction of the involved controllers with the extended system and its newly developed functionalities (Table 2). For the numbers in parentheses, the value 1 represents the answer "strongly disagree", the value 4 represents "neutral", and the value 7 represents "strongly agree", with the other values in between assigned accordingly. The ATCOs exhibited high agreement with the notion that an AMAN can be operationally improved by integrating adverse weather information (6.4/7) and adverse weather guidance support (6/7). They also showed a high level of agreement with the concept that an extended AMAN would lead to improved decision making (6/7). Furthermore, they reported their agreement with the statement suggesting that they would feel comfortable using the visual aids on the radar display (6/7). All in all, they found the proposed concept for extended Arrival Management very useful (6/7).

**Table 2.** Detailed overview of the results of the questionnaires with graded answer options.


In summary, the evaluation of the validation with controllers showed that they would very much like to use a system that supports them in approach planning and flying around severe weather areas, both from a planning and visual point of view, as it would make their daily work easier and thus also safer. It is noteworthy that they placed more emphasis on weather information than on planning support.
