Skip to main content
European Commission logo
español español
CORDIS - Resultados de investigaciones de la UE
CORDIS
CORDIS Web 30th anniversary CORDIS Web 30th anniversary

Artificial Neural Networks for the Prediction of Contrails and Aviation Induced Cloudiness

Periodic Reporting for period 1 - E-CONTRAIL (Artificial Neural Networks for the Prediction of Contrails and Aviation Induced Cloudiness)

Período documentado: 2023-06-01 hasta 2024-05-31

The overall purpose of E-CONTRAIL project is to develop artificial neural networks (leveraging remote sensing detection methods) for the prediction of the climate impact derived from contrails and aviation-induced cloudiness, contributing, thus, to a better understanding of the non-CO2 impact of aviation on global warming and reducing their associated uncertainties as essential steps towards green aviation.

Specifically, the objectives of E-CONTRAIL are:
• O-1 to develop remote sensing algorithms for the detection of contrails and aviation-induced cloudiness.
• O-2 to quantify the radiative forcing of ice clouds based on remote sensing and radiative transfer methods.
• O-3 to use of deep learning architectures to generate AI models capable of predicting the radiative forcing of contrails based on data-archive numerical weather forecasts and historical traffic
• O-4 to assess the climate impact and develop a visualization tool in a dashboard

Upon successful achievement of the objectives described above, we ambition to provide aviation stakeholders with an early and accurate (thus, reducing the associated uncertainty) prediction of those volumes of airspace with the conditions for large global warming impact due to contrails and aviation-induced cloudiness
a. Technical work linked to objective 1.
The technical work to be carried out to achieve objective 1 is included in WP 1. WP 1 is completed (deliverable D1.1 has been approved with some minor comments by SJU and re-submitted, D1.2 has been submitted; (MS-4 achieved upon approval of D1.2). Tasks 1.1 Task 1.2 and Task 1.3 have been completed. Therefore, the target achievement is 100%. See overall assessment in Section 1.2.1.

b. Technical work linked to objective 2.
The technical work to be carried out to achieve objective 2 is included in WP 2. WP 2 is in progress (deliverables D2.1 and D2.2 have been approved by SJU). Tasks 2.1 and Task 2.2 have been completed. Task 2.3 is 50% fulfilled; Task 2.4 is 30% fulfilled. Hence, the achievement of the objective is at 70 %. Risk 5 is active. See overall assessment in Section 1.2.2.

c. Technical work linked to objective 3.
The technical work to be carried out to achieve objective 3 is included in WP 3. WP 3 is in progress (deliverable D3.1 has been approved by SJU). Task 3.1 is at 100% fulfilled, Task 3.2 at 10% fulfilled, and Task 3.3 at 0% fulfilled. Consequently, the achievement of the objective is at 50%. See overall assessment in Section 1.2.3.

d. Technical work linked to objective 4.
The technical work to be carried out to achieve objective 4 is included in WP 4. WP 4 is in progress (deliverable D4.1 and D4.2 has been submitted). Task 4.1 is completed at 80% fulfilled. Task 4.2 is completed at 10%. Task 4.3 is at 5%. All in all, the achievement of the objective is at 40%. See overall assessment in Section 1.2.4


Project achievement (overall assessment)

At the end of this 1st Reporting Period, the overall achievement can be set at about 40%.
Due to the delay in receiving the MTG data, some of the objectives have been delayed beyond expectations. Objectives described above are progressing at good pace:
• Objective 1 – WP1 is at 100% fulfilled.
• Objective 2 – WP2 is at 70% fulfilled.
• Objective 3 – WP4 is at 50% fulfilled.
• Objective 4 – WP5 is at 40% fulfilled.
We have successfully developed a model capable of detecting contrails in satellite imagery and another for characterizing the evolution of contrails into aviation-induced cloudiness. A nearly 10% improvement in the Dice score for contrail detection models was achieved. Necessary pre- and post-processing steps for optimal performance were outlined. The models were demonstrated to be applicable to European data, though further quantitative validation with MSG/SEVIRI images is pending. Performance is expected to improve with the availability of MTG/FCI data.

We have developed a physics-driven contrail simulation model, including a novel transport equation to address slip mechanisms, simulating the advection-diffusion of ice particles. Persistent contrail propagation is governed by a comprehensive advection-diffusion equation, essential for precise simulations and radiative forcing assessments. While the model requires further refinement, it shows reasonable consistency with methods like CoCiP. Future enhancements will focus on the: 1) Physics Representation: Updating the model to incorporate additional physical processes; 2) Large-Scale Simulation: Reducing computational time for tracking contrails over extensive areas such as Europe, improving algorithm efficiency. The work conducted under this task lays the groundwork for future advancements in understanding and mitigating aviation's climate impact through improved detection and simulation techniques.