Invitation: Join our new IARAI Multi-sensor Weather Forecast Competition
Posted: Fri Apr 02, 2021 1:33 am
Join our new Multi-sensor Weather Forecast Competition!
Study multi-channel weather movies.
Predict weather products in various Earth regions.
Apply transfer learning to new Earth regions.
Goal
The goal of the competition is a short-term prediction of selected weather products in three regions (core challenge) and apply the transfer learning to predict weather products in the three additional regions (transfer learning challenge).
Following our recent success of our Traffic4cast competitions at NeurIPS in 2019 and 2020, this challenge similarly presents the weather forecast as a video frame prediction task. The competition goal is to predict the next 32 images (8 hours in 15-minute intervals) of the weather movies. The images contain four channels encoding the following weather products: temperature (on the accessible surface: top cloud or earth), convective rainfall rate, probability of occurrence of tropopause folding, and cloud mask, based on meteorological satellites data obtained in collaboration with AEMET/ NWC SAF. Each pixel in the images represents the area of about 4 km x 4 km, and each region contains 256 x 256 pixels. The regions span varying landscapes including mountains, deserts, islands and seas, and others
Join us and learn more about the data on our website: weather4cast.ai
Prizes
The winners of the core competition and the transfer learning competition will be awarded the following prizes:
1st place – a voucher or cash prize worth 5,000€ to the participant/team;
2nd place – a voucher or cash prize worth 3,000€ to the participant/team;
3rd place – a voucher or cash prize worth 2,000€ to the participant/team.
Deadlines
Competition starts: April 1, 2021
Submission deadline: May 31, 2021, 23:59 AOE.
held-out dataset available: June 1, 2021, 00:00.
Abstract submission and held-out dataset prediction: June 6, 2021, 23:59 AOE.
Announcement of the winners: June 21, 2021.
Study multi-channel weather movies.
Predict weather products in various Earth regions.
Apply transfer learning to new Earth regions.
Goal
The goal of the competition is a short-term prediction of selected weather products in three regions (core challenge) and apply the transfer learning to predict weather products in the three additional regions (transfer learning challenge).
Following our recent success of our Traffic4cast competitions at NeurIPS in 2019 and 2020, this challenge similarly presents the weather forecast as a video frame prediction task. The competition goal is to predict the next 32 images (8 hours in 15-minute intervals) of the weather movies. The images contain four channels encoding the following weather products: temperature (on the accessible surface: top cloud or earth), convective rainfall rate, probability of occurrence of tropopause folding, and cloud mask, based on meteorological satellites data obtained in collaboration with AEMET/ NWC SAF. Each pixel in the images represents the area of about 4 km x 4 km, and each region contains 256 x 256 pixels. The regions span varying landscapes including mountains, deserts, islands and seas, and others
Join us and learn more about the data on our website: weather4cast.ai
Prizes
The winners of the core competition and the transfer learning competition will be awarded the following prizes:
1st place – a voucher or cash prize worth 5,000€ to the participant/team;
2nd place – a voucher or cash prize worth 3,000€ to the participant/team;
3rd place – a voucher or cash prize worth 2,000€ to the participant/team.
Deadlines
Competition starts: April 1, 2021
Submission deadline: May 31, 2021, 23:59 AOE.
held-out dataset available: June 1, 2021, 00:00.
Abstract submission and held-out dataset prediction: June 6, 2021, 23:59 AOE.
Announcement of the winners: June 21, 2021.