Welcome 👋

Reproducibility Challenge

Welcome to the 2023 Climate Informatics Reproducibility Challenge!

Objectives

This year’s challenge aims to build community, facilitate collaboration, and advance open science within the Climate Informatics community, continuing the tradition of an annual Climate Informatics conference-associated hackathon whilst drawing inspiration from The ML Reproducibility Challenge.

Overview

In this year’s challenge, teams of 2-4 will collaborate to create a notebook which reproduces the key contributions of a published environmental data science paper for eventual integration in the open-source Environmental Data Science (EDS) Book. Over the course of a month, teams will locate the data and code associated with their chosen paper; train, validate, and test the models used in the paper; visualise key results from these experiments; discuss their work with peer reviewer(s); and ultimately weave together a narrative illuminating the value of open science which culminates in a citeable, DOI-tagged notebook. Teams will further have the opportunity to network and exchange technical Q&A with fellow participants in weekly drop-in socials throughout the competition. Check out this notebook in the EDS book which reproduces IceNet (Andersson, 2021) for an example of what you’ll be working towards!

Cloud infrastructure

To facilitate the reproducibility of papers, a cloud infrastructure with a Pangeo Machine Learning JupyterHub is being made available (registration to this service is required and free) by the European Open Science Cloud (EOSC).

Information about access to the JupyterHub can be found here.

Tip

The computing and storage resources used for this challenge have been kindly provided by CESNET in the context of the C-SCALE project that provides advanced computing for EOSC.

	The C-SCALE project (Copernicus – eoSC AnaLytics Engine) has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 101017529.

Summary

Too long; didn’t read? Here is the TL;DR on the 2023 Climate Informatics Reproducibility Challenge:

• Teams of 2-4

• free Pangeo ML jupyterhub available on EOSC C-SCALE cloud infrastructure

• Reproducing a pre-approved environmental data science or participant-suggested paper as a notebook

• Submissions to be judged after one month

• Winning team to receive Cambridge University Press book of their choosing (up to £500 in value split across the winning team)

• Authorship on a citeable, DOI-tagged notebook to be integrated within the Environmental Data Science (EDS) Book

• Great networking and community-building opportunity

Organising Committee

The 2023 Climate Informatics Reproducibility Challenge Organising Committee warmly welcomes your participation.

• Andrew McDonald, University of Cambridge and British Antarctic Survey

• Alejandro Coca-Castro, The Alan Turing Institute

• Anne Fouilloux, Simula Research Laboratory

• Andrew Hyde, Cambridge University Press

Contact

Please direct any questions to Andrew McDonald at arm99@cam.ac.uk. Follow us on Twitter @Climformatics for updates as the conference and competition approaches!

Acknowledgments 🙌

2023 Climate Informatics Reproducibility Challenge is co-hosted by EDS book, Climate Informatics and Cambridge University Press & Assessment with support from Cambridge University, The Alan Turing Institute and Simula Research Laboratory.

Cloud computing infrastructure is kindly provided by C-SCALE.