Redefining decision making: introducing probabilistic forecast products to aviation applications

There are no aviation operations without reliable weather information. Efficient and safe air traffic management relies on accurate meteorological predictions on different timescales from nowcasting to the midrange. On top of that, it is crucial to enable a safe interpretation of uncertain weather data so that these forecasts are fruitful for planning and decision making within aircraft operations.

So far, the aviation meteogram product issued by MeteoSwiss consists of deterministic predictions and threshold exceedance probabilities. This does not exploit the full potential of the underlying forecasts, as (1) deterministic predictions may be biased, and (2) the current product does not present the full uncertainty picture to decision makers.

In response to these challenges, we propose a transition towards delivering probabilistic forecasts. This shift unlocks the full potential of information-based decision making, which finally allows smoother and  economically and ecologically more sustainable aviation operations. Importantly, as probabilistic data and its potential is largely unknown to our customers so far, a robust program of frequent training and education is necessary.

In this presentation, we showcase new machine learning ensemble predictions for thunderstorms, wind, and visibility at airports in Switzerland for the nowcasting and short-term forecasting range. In particular, we focus on their comprehensive visualization in a meteogram tailored to convey ensemble information and discuss challenges, advantages and future ideas.