[Webinar Series] CONTRAILS, Climate, and AI-Driven Observation

[Webinar Series] CONTRAILS, Climate, and AI-Driven Observation: Research Insights and Future Directions

The formation, persistence, and radiative impact of contrails depend on a complex interplay of atmospheric conditions, aviation operations, and cloud microphysics.The CONTRAILS project aims to provide tools to monitor the impact of contrails on climate. This webinar concludes the CONTRAILS project – a joint work of the German Weather Service DWD (DE), Thales Research & Technology (FR), Reuniwatt SAS (FR) and Laboratoire Atmosphères, Milieux, Observations Spatiales LATMOS (FR) – and situates its scientific results within the broader European contrails research and development landscape. It brings together project partners and experts to review key outcomes of CONTRAILS, discuss methodological advances, and highlight how ongoing and upcoming initiatives are building on these results.

In this webinar, speakers from Reuniwatt, Thales, and LATMOS will present the final scientific and technical results of the project, with a particular focus on observation strategies and AI-driven analysis methods. The session will show how visible and infrared all-sky imagers, lidar observations, and satellite-based cloud measurements can be used together to capture contrail occurrence, altitude, dynamics, and evolution. It will also explain how AI methods support contrail detection, clustering, stereoscopic reconstruction, dynamics prediction, and the linkage between observations and physical models.

The session will open with an overview of European contrails research, positioning CONTRAILS within the wider ecosystem of national and international projects. This introduction will be followed by in-depth presentations of the project’s scientific work, before opening the discussion to the broader contrails community.

LATMOS will present the atmospheric and climate-science perspective, including the role of lidar measurements, satellite observations, and radiative-impact analysis. Particular attention will be given to how lidar can provide vertical information on cirrus and contrail-related cloud structures, and how satellite observations such as IASI and CALIPSO can support the evaluation of contrail radiative forcing and the validation of model parameterisations.

Thales will present advances in AI methods for contrail identification and dynamics prediction, including approaches based on geometric deep learning and physics-informed AI. These methods are designed to process fisheye sky-imager data more naturally, improve robustness against image distortion, and combine data-driven learning with physical knowledge. This part of the webinar will address how AI can move beyond simple image recognition toward the interpretation of contrail formation, motion, persistence, and potential climate relevance.

Finally, the webinar will highlight Reuniwatt’s contribution to the deployment and use of all-sky imaging technologies for contrail monitoring, including data acquisition, calibration, annotation, and the creation of datasets suitable for AI-based processing. It will also cover the use of visible and infrared imagery for the analysis of contrail structures and stereoscopic observation approaches, extending all-sky camera methods from cloud monitoring toward contrail-specific climate research.

The session will be chaired by Teodora Petrisor, who will provide an overview of the European contrails R&D landscape and position the CONTRAILS project within ongoing research on aviation’s non-CO₂ climate effects. She will introduce the project’s objectives, scope, and the methodological link between observation, AI-based analysis, physical modelling, and climate-impact assessment.

Philippe Keckhut will present the LATMOS contribution to the project, focusing on the atmospheric-science perspective, lidar-based cloud observations, satellite data, and the interpretation of contrails within cirrus-cloud systems.

Davide Di Giusto and Solène Blasco-Lopez will present Thales’ work on AI-based contrail identification and dynamics prediction, including the use of image analysis and learning algorithms to detect contrails and follow their evolution from ground-based observations. This part will explain how AI can help transform complex sky-imager data into operationally meaningful information for contrail monitoring, and how these methods aim to improve robustness, interpretability, and the coupling between data-driven learning and physical atmospheric models.

Olivier Liandrat will present Reuniwatt’s contribution to contrail observation technologies, with a focus on visible and infrared all-sky imagers, observation campaigns, calibration, data acquisition, and stereoscopic approaches. His contribution will show how ground-based imaging supports the creation of annotated datasets and the validation of AI-based contrail analysis.

After this webinar, you will have a better understanding of:

About the speakers:

Dr. Teodora Petrisor is a senior research engineer in AI at CortAIx Labs, Thales, leading the AI algorithm validation roadmap. She holds a PhD in Signal and Image Processing from Telecom Paris. For the past couple of years she has been focusing on trustworthy AI algorithms with applications to green operations in avionics and more specifically to contrail detection and prediction. She is currently leading the French consortium in the BPI/DLR CONTRAILS project.

Prof. Dr. Philippe Keckhut is professor at the University of Versailles Saint-Quentin-en-Yvelines (UVSQ). He obtained his PhD from Sorbonne University on Atmospheric Physics in 1991, after its master 2 in Optics. He has published more than 190 peer-reviewed articles on lidar techniques, atmospheric physico-chemistry, cirrus clouds, and decadal atmospheric monitoring. He has led the French contribution of the international stratospheric Ozone network (NDACC. the Network of Detection Atmospheric Composition Changes). He has co-created the company GORDIEN_STRATO, that developed Lidars technologies. Visiting researcher at NOAA, he worked on space observations using SSU/NOAA, and UARS/NASA satellites, then being involved on ENVISAT/ESA platform, has contributed to a series of 3 nano-satellites, launched from 2021 to 2025, and has created a master 2 diploma named Newspace.

He was vice-director of the IPSL consortium (FR 636) in charge of observations, vice-director of the UVSQ Observatory (UMS 3342), director of LATMOS (UMR 8190), vice-president Research and Innovation at UVSQ and he is now director of the France 2030 project called Académie Spatiale d’Île-de-France. His work has been awarded through the Louis Armand Prize from the Academy of Sciences, by the Michel Cépède Medal from the Society for Research and Invention, by the Norbert-Gerbier Mumn Prize from the World Meteorological Organization, and was named as Knight of the Order of Academic Palms.

Olivier Liandrat is an engineer in Computer Science and Applied Mathematics (Grenoble INP-Ensimag) specialised in image processing. He joined Reuniwatt in 2012 for his end-of-study internship in order to conduct the company’s first tests on short-term irradiance prediction from wide-angle images of the sky. As the company’s first employee, he actively participates in Reuniwatt’s growth, and contributed to the first patent filing of Reuniwatt’s Sky InSight™™ wide-angle thermal camera. At the end of 2015, he joined the IRT Saint-Exupéry to carry Reuniwatt’s expertise within the ALBS research project. Alongside Airbus DS, Thales Alenia Space and CNES, the project developed the essential building blocks to adapt the Sky InSight™™ to the challenges of laser optical communication. Building on this experience, he is now overseeing the company’s further developments for cloud-sensitive markets, including air traffic management.

Solène Blasco-Lopez is a research engineer in AI and computer vision at Thales, where she works on machine learning methods for image and video analysis across multiple modalities, including RGB, infrared, and satellite data. Solène holds a Master’s degree in Mathematics, Vision, and Learning from ENS Paris‑Saclay, and an engineering degree from ENSAE Paris, with a specialization in statistics, machine learning, and data science.

Davide Di Giusto, PhD, is a research engineer at Thales CortAIx Labs France, where he works on Physics-Informed AI solutions for aeronautical applications. Before that, he obtained his PhD in fluid mechanics from in a joint degree between the Aix-Marseille Université (FR) and the Università di Udine (IT), working on both high-performance simulations and experimental approaches to better characterize the dispersion of micro-plastics in fluid flows.

Teodora Petrisor

Moderator
Senior Research Engineer AI @Thales

Prof. Philippe Keckhut

Speaker
Lecturer at UVSQ / Director of the Île-de-France Space Academy project @LATMOS

Olivier Liandrat

Speaker
Sky Imaging Team Lead @Reuniwatt

Solène Blasco-Lopez

Moderator
Research Engineer AI & Computer Vision @Thales

Davide Di Giusto

Speaker
Research Engineer Software @Thales