AI weather

A smarter forecasting

The need for improved weather information has already been acknowledged by the international aviation community. Efforts by the Meteorology Panel of the International Civil Aviation Organization (ICAO) are already working to deliver the next generation of the World Area Forecast System (WAFS) by the end of the current decade. The next generation of WAFS information will provide increased spatial and temporal resolution, increase the frequency release time, incorporate probabilistic forecasts, and be provided in the new IWXXM format established by ICAO.


MLP and LSTM neural networks for storm prediction

Thunderstorms are frequent in the summer and coincide with a period of high air-traffic demand in the European airspace. This combination of bad weather and high demand causes significant disruption to air-traffic-management operations. Artificial intelligence has great potential to automate and improve the quality of meteorological data available to the aviation community. By providing a constant feed of forecasts and observation data, the models are able to learn in a continuous manner to predict with greater accuracy the occurrence of disruptive weather events.

Development of decision support tools from AI model results

A key aspect in moving to an industrial application is developing visualization techniques for displaying model results. In this regard, operational feedback from end-users is of paramount importance. A prototype of the AI-based convection forecast product was produced representations using the established 5-level risk matrix used within the EUROCONTROL Cross-Border Weather initiative by UC3M spin-off, INTELMET. The resulting color-coded georeferenced polygons allow air traffic managers to visualize various threat levels from convective weather.

PINNs to enhance existing atmospheric sensor data

The novel development of Physics-Informed Neural Networks (PINNs), which incorporate the constraints given by physics laws into the training process, has opened many possibilities to enhance experimental data. Their combined implementation together with Generative Adversarial Networks (GANs) – robust networks with excellent precision but excessive computational costs – enables the development of interesting applications regarding the forecast of severe weather conditions, where PINNs are useful to compute the fluidic behavior of storms approaching a certain location, whereas GANs can incorporate many additional parameters, such as wind speed, humidity, temperature and electric content.

Related projects

  • ISOBAR ↗ Artificial Intelligence Solutions to Meteo-Based DCB Imbalances for Network Operations Planning.
  • ALARM ↗ A prototype of a global multi-hazard monitoring and early warning system for natural hazards in aviation.
  • MetATS Managing meteorlogical uncertainty for a more efficient air traffic system.
  • StormATS– Managing Storm Uncertainties for a Safe and Efficient Air Transport System.
  • INTELMET PoC – Improving the lead time and accuracy of thunderstorm predictions by utilising artificial intelligence.
  • PERSEVERE – Physics-informed nEuRal networks for SEVERe wEather event prediction

Highlighted publications

  • Chkeir, Sandy and Anesiadou, Aikaterini and Cervantes, Alejandro and Reviriego, Alvaro and Soler, Manuel and Biondi, Riccardo, A New Extreme Weather Nowcasting Product Supporting Aviation Management at Local Scale. Under Review in Weather and climate extremes. In pre-print available at SSRN: or
  • Aniel Jardines, Manuel Soler, Javier García-Heras, Laure Raynaud, Matteo Ponzano. Pre-tactical Convection Prediction for Air Traffic Flow Management using LSTM Neural Network. Journal of Air Transport Management.
  • Iván Martínez, Aniel Jardines, Manuel Soler, Alejandro Cervantes, Javier García-Heras. Predicting Air Traffic Flow Management Regulations due to Weather using Convolutional Neural Networks–  Transportation Research Part C.
  • Jardines, Aniel and Eivazi, Hamidreza and Zea, Elías and Simarro, Juan and García-Heras, Javier and Soler, Manuel and Otero, Evelyn and Vinuesa, Ricardo, Thunderstorm Prediction During Pre-Tactical Air-Traffic-Flow Management Using Convolutional Neural Networks. Expert Systems with Applications. Under Review Available at SSRN:
  • Estimating Entry Counts and ATFM Regulations during Adverse Weather Conditions using Machine Learning. Aniel Jardines, Manuel Soler, Javier García-Heras. Journal of Air Traffic Management, 2021
  • Convection Indicator for Pre-Tactical Air Traffic Flow Management using Neural Network. Aniel Jardines, Alejandro Cervantes, Manuel Soler, Javier García-Heras, and Juan Simarro.  Machine Learning with Applications. 2021-06 | journal-article DOI: 10.1016/j.mlwa.2021.100053