TBO-Met (Meteorological Uncertainty Managementfor Trajectory Based Operations)A SESAR H2020 Exploratory Research Project Funded by

TBO-Met ConsortiumTBO-Met Consortium: UNIVERSITY OF SEVILLE (USE, Coordinator), METEOSOLUTIONS GmbH (MetSol), PARIS-LODRON-UNIVERSITAT SALZBURG(PLUS), AEMET, and UNIVERSIDAD CARLOS III DE MADRID (UC3M)Contact: Damian Rivas (USE, Project coordinator), Juergen Lang (MetSol), Carl-Herbert Rokitansky (PLUS), Juan Simarro (AEMET), and Manuel Soler (UC3M)Website: https://tbomet-h2020.com

Abstract

In TBO-Met project the problem of analysing and quantifying theeffects of meteorological uncertainty in Trajectory Based Operationsis addressed. In particular, two problems are considered: 1) trajec-tory planning and 2) sector demand analysis, both at the pre-tacticallevel (up to three hours before departure) and tactical level (duringthe flight). In each problem two types of meteorological uncertaintyare considered: wind uncertainty and convective zones (including in-dividual storm cells). Weather predictions will be based on EnsemblePrediction Systems and Nowcasts. At the trajectory scale, the mainobjective is to assess and improve the predictability of efficient 4D tra-jectories when weather uncertainty is taken into account. To reach thisgoal, a methodology based on the use of stochastic optimal controlalgorithms will be explored for robust trajectory planning at the pre-tactical level. At the tactical level, various tactics will be investigatedto avoid storms by using a Monte-Carlo method. At the sector scale,the main objective is to analyse the impact of the previously devel-oped trajectory planning on sector demand. To achieve this objective,a methodology will be developed to measure the uncertainty of sectordemand (probabilistic sector loading) based on the uncertainty of theindividual trajectories. This analysis will also provide an understand-ing of how weather uncertainty propagates from the trajectory scale tothe sector scale. All solutions proposed in this project will be evaluatedand assessed using an advanced air traffic simulator.

Introduction

A better understanding of the elements introducing uncertaintyin the traffic is key when optimizing, planning, executing,monitoring and synchronizing trajectories with ground systemsand/or other aircraft. In particular, the need of computing ef-ficient, yet accurate trajectories becomes the fundamental cor-nerstone for reaching the expected benefits of TBO in terms ofincreased capacity, increased efficiency, and reduced environ-mental impact.

The analysis of uncertainty in ATM must take into accountthe time horizon and the different scales of the system. Whilethe spatial uncertainty affects mainly safety issues (loss of sep-aration) and efficiency, the temporal uncertainty manifests itselfprimarily as delay. Three scales of the system can be clearlydifferentiated: 1) microscale (a single flight); 2) mesoscale (airtraffic); and 3) macroscale (the air transport network).

Weather uncertainty is one of the main sources of uncertaintythat affect the ATM system. In this project we focus on the anal-ysis of meteorological uncertainty coming from the followingtwo sources: 1) wind, and 2) convective regions, including indi-vidual storm cells.

Main Objectives

The overall objective of the project is threefold:

1. To advance in the understanding of the effects of meteorolog-ical uncertainty in TBO.

2. To develop methodologies to quantify and reduce the effectsof meteorological uncertainty in TBO.

3. To pave the road for a future integration of the managementof meteorological uncertainty into the air traffic managementsystem.

The analysis of the effects of meteorological uncertainty in TBOis an extraordinarily broad task. In this project we focus on twoparticular problems, both at the pre-tactical and tactical levels:Trajectory planning and prediction of sector demand. Thesetwo problems correspond to two scales of the ATM system: air-craft trajectory (microscale) and air traffic in en-route sectors(mesoscale). Hence, in this project there are two particular ob-jectives:

• To improve the predictability of aircraft trajectories when sub-ject to meteorological uncertainty keeping acceptable levelsof efficiency, which is to be accomplished by developing amethodology to quantify the trade-off between predictabilityand efficiency.

• To increase the accuracy of the prediction of sector demandwhen meteorological uncertainty is taken into account, whichwill be achieved by developing a methodology to measure theimpact of improved trajectory planning under meteorologicaluncertainty on sector demand, that is, by quantifying the im-provement of the predictability of the sector demand when thepredictability of the individual trajectories is improved.

WP Structure

Figure 1: Work-Package Structure

Progress Results

WP2

According to the requirements specified to the research topicsof WP4 and WP5, the needed meteorological data sources, i.e.,suitable ensemble prediction systems have been identified andthe necessary processing methods defined. A spatial-temporalgrid will be built: the processing covers coordinate transforma-tion from hybrid model levels to pressure levels, vertical interpo-lation, temporal downscaling and interpolation, spatial bilinearinterpolation and the extraction of polygons which delimit ar-eas of e.g. deep convection. Further data processing is definedin order to calculate ensemble mean and spread of wind com-ponents and temperature which is used to quantify the forecastuncertainty of these meteorological parameters. While wind andtemperature data is readily available as model output, informa-tion about convection can be derived from numerous parameters.Suitable indicators to describe convection have been identified.

Figure 2: ECMWF-EPS and GLAM-EPS data and processing diagrams forwind, temperature, and convection.

WP3

A TBO-Met Survey Questionnaire has been prepared. The fun-damental objectives of the survey are twofold: 1) ensure thatTBO- Met project is aligned with current meteorological prac-tices in aviation (particularly any issue regarding meteorologicaluncertainty); and 2) understand future expectations and needsregarding meteorological uncertainty management.

WP4

We present preliminary results on robust trajectory planning atpre-tactical level. The main goal is to predict trajectories that areefficient, yet predictable. State of the art Ensemble ProbabilisticForecasts are used as data for wind (assumed to be the uniquesource of uncertainty). An ad-hoc optimal control methodol-ogy has been developed to solve trajectory planning problemsconsidering wind input from EPFs. A set of Pareto Optimal tra-jectories is obtained, in particular results for the minimum fueltrajectory and the most predictable trajectory are presented anddiscussed. Trade-off between fuel consumption and time disper-sion is obtained. It is shown how uncertainty can be quantifiedand reduced by proposing alternative trajectories. Please, referto [1] for further information.

100◦W

80◦W

60◦W 40◦W 20◦W 0◦

30◦N

40◦N

50◦N

60◦N 60◦N

0.0

1.5

3.0

4.5

6.0

7.5

9.0

10.5

pσ2 u+σ2 v

p

29000 29500 30000 30500 31000

Fuel consumption (kg)

0.5

1.5

2.5

3.5

4.5

Arrivaltimerange(minutes)

p=0

p=50

p

Figure 3: up: Optimal trajectories from NY to Lisbon, for values of p (pre-dictability parameter) from 0 (min. fuel) to 50 (max. predictability). Higherbrightness in the trajectory color indicates higher values of p. We also colorregions of higher uncertainty, which we have defined as

√σ2u + σ2v, with σu

being the standard deviation of the u component of wind across differentmembers and σv analogous for the v-component. down: Pareto frontier ofthe problem

0 1000 2000 3000 4000 5000 6000−150

−100

−50

0

50

100

150

200

Tim

ele

ador

lag

(s)

0 1000 2000 3000 4000 5000 6000250

260

270

280

290

300

310

Gro

und

spee

d(m

/s)

0 1000 2000 3000 4000 5000 6000Distance (km)

1.4

1.5

1.6

1.7

1.8

1.9

2.0

Hea

ding

(rad

)

CourseTrue heading

0 1000 2000 3000 4000 5000 6000 7000−150

−100

−50

0

50

100

150

200

Tim

ele

ador

lag

(s)

0 1000 2000 3000 4000 5000 6000 7000250

260

270

280

290

300

310

Gro

und

spee

d(m

/s)

0 1000 2000 3000 4000 5000 6000 7000Distance (km)

0.5

1.0

1.5

2.0

2.5

3.0

Hea

ding

(rad

)

CourseTrue heading

Figure 4: State-space evolution in the case p = 0 and p = 50 (notice that thesolution to the problem provides a unique course, and a unique true airspeed[set as a constant parameter], both to be tracked. In addition, it providesdifferent tracks and ground speeds associated to the different winds in the en-samble). Time leads or lags are defined with respect to the average trajectory.Left: min. fuel (p=0); Right: max. predictability (p=50)

References[1] D. Gonzalez-Arribas, M. Soler, and M. Sanjurjo, “Wind-

based robust trajectory optimization using meteorologi-cal ensemble probabilistic forecasts,” in SESAR InnovationDays, 2016.

AcknowledgementsTBO-MET project has received funding from the SESARJU under grant agreement No 699294 under Euro-pean Union’s Horizon 2020 research and innovation pro-gramme. Consortium members are UNIVERSITY OFSEVILLE (Coordinator), AEMET (Agencia Espanola deMeteorologıa), METEOSOLUTIONS GmbH, and PARIS-LODRON-UNIVERSITAT SALZBURG, and UNIVERSITYCARLOS III DE MADRID. Website: https://tbomet-h2020.com