The provision of weather updates to air traffic controllers and pilots is a critical element within the SESAR vision of increasing flight predictability and the overall performance of the air traffic management (ATM) system. To support this vision, IMET, a recently-completed SESAR exploratory research project under WP-E, assessed a number of techniques to reduce weather forecast uncertainty on trajectory predication.

The IMET project aimed to investigate, and preferably provide recommendations, how to use the uncertainty information from Ensemble Weather Forecast (EWF) to optimize TP output with respect to the EWF, and hence the future weather, as it is assumed the EWF covers all possible futures.

Jointly carried out by Netherlands Aerospace Center NLR, and national weather service providers (Météo France and UK Met Office), the 30-month project assessed the level of sensitivity of several key aircraft trajectory parameters to meteorological forecast uncertainty.

The IMET approach involves the use of a whole set or “ensemble weather forecast”, rather than one weather forecast, based on the output of a numerical weather prediction system using advanced atmospheric models and a range of observations. It is generally understood that no single member can predict the weather perfectly, but the weather ensemble is constructed in such a way that it should cover all possible futures.

One parameter that is impacted by weather forecasts is the amount of fuel needed for a flight. For long-haul flights, a contingency of fuel is calculated for taking into account possible additional en-route fuel consumption caused by weather, routing changes or ATM restrictions. Many airlines determine its actual value using a single deterministic weather forecast, a trajectory predictor (TP), and by comparing predicted weather conditions with previously experienced meteorological conditions for the same route.

In the IMET approach, for each ensemble member, the TP computes the (amount of) contingency fuel. The resulting set of predicted fuel values provides the information to estimate the uncertainty of the contingency fuel due to weather. If the fuel uncertainty is small, which is usually the case if the forecast weather conditions are excellent or good to conduct flight operations, the predicted mean fuel value resembles the value computed with the single weather forecast. However, if the uncertainty of the predicted fuel value is big, which is often the case in disruptive weather conditions, choosing the value computed with the single weather forecast could be significantly inaccurate. This in turn can lead to an underestimated fuel contingency and the need for refuelling. On the other hand, if the actual weather appears to be significantly better than expected, the flight cost may turn out higher than necessary due to the carrying of unused fuel. By translating weather uncertainty to fuel uncertainty in this way allows for more effective decision making.

In IMET, several techniques were explored to reduce the uncertainty in the weather forecast for trajectory prediction, particularly regarding the expected weather condition and its location in space and time. Among these techniques are “forecast bias correction”, which aims to improve the accuracy of the weather forecast by removing small scale levels of uncertainty and “noise” from the observation data (see Figure 1 ), and “upscaling of the meteorological data”, a method to reduce the uncertainty of forecast 4D location errors at the expense of accuracy.

When applied, the IMET approach could bring positive benefits in terms of predictability by quantifying the integrated trajectory uncertainty parameters due to weather. IMET may also contribute to cost reduction, due to reduced contingency fuel usage if the ensemble weather forecast is capable of identifying the weather conditions with a high level of confidence.

The results of the project have being taken up by the SESAR industrial research project on flight planning (SESAR WP11.1).

Example of smoothing of observed maximum radar reflectivity data from French weather radar system ARAMIS at 2014-06-09 18 UTC: raw data (left), and smoothed (right)