Challenge

In a nutshell:

Weather4cast 2023 is based on the same unique data set underpinning last year’s competition at NeurIPS with the new challenge of not just predicting binary rainfall masks but actual rainfall amounts.

At the same time, the competition unexpectedly had to change its host, delaying the originally planned exploratory Phase-I.

As a result, we are directly diving into the main competition phase, and will leave the exploratory discussions to our NeurIPS Competition Track session to which all delegates are cordially invited to contribute.

Prediction task
Competition participants should predict the exact amount of rainfall for the next 4 hours in 32 time slots from an input sequence of 4 time slots of the preceeding hour. The input sequence consists of four 11-band spectral satellite images. These 11 channels show slightly noisy satellite radiances covering so-called visible (VIS), water vapor (WV), and infrared (IR) bands. Each satellite image covers a 15 minute period and its pixels correspond to a spatial area of about 12km x 12km. The prediction output is a sequence of 32 images representing rain rates from ground-radar reflectivities. Output images also have a temporal resolution of 15 minutes but have higher spatial resolution, with each pixel corresponding to a spatial area of about 2km x 2km. So in addition to predicting the weather in the future, converting satellite inputs to ground-radar outputs, this adds a super-resolution task due to the coarser spatial resolution of the satellite data

You can now get the data, download our Starting Kit from github and start training your first model! The competition leaderboard is coming soon!

We will jointly publish NeurIPS Databased & Benchmarks Proceedings article this year. We are now working with NeurIPS organizers to receive free registrations for the winning submissions.

Leaderboards