Home

Welcome!   This page will provide challenge problems based on the SEVIR weather dataset. 

Currently, we have an initial release of the nowcasting challenge, which includes a sample training/testing dataset, baseline model, and evaluation metrics.  See the notebook in this repo:

https://github.com/MIT-AI-Accelerator/sevir_challenges/tree/main/radar_nowcasting

Overview

Modern deep learning approaches have shown promising results in meteorological applications like precipitation nowcasting, synthetic radar generation, front detection and several others. We are machine learning researchers working collaboratively across MIT, MIT Lincoln Laboratory, and the Air Force to rapidly develop new approaches to these challenges. We are seeking collaborators from across the globe to leverage the "machine learning ready" dataset we have curated. Check out the SEVIR dataset for more details.

Challenge Dataset

The dataset used for this challenge will be the Storm EVent ImageRy (SEVIR) dataset.   Click on the data tab for more information.

Challenge Tasks

This challenge will consist of the following tasks (additional tasks may be added in the future, so continue to check)

  1. Nowcasting

Submission Guidelines

To submit a new entry to the leaderboard, please add your result evaluated on the provided test set to the leader board and add a reference to a paper describing your method.  Submit as a pull request to https://github.com/MIT-AI-Accelerator/sevir_challenges

 

 

-------------------

Research was sponsored by the United States Air Force Research Laboratory and the United States Air Force Artificial Intelligence Accelerator and was accomplished under Cooperative Agreement Number FA8750-19-2-1000. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the United States Air Force or the U.S. Government. The U.S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation herein.