The Climate Analytics Lab aims to reduce uncertainty in the role of anthropogenic aerosol on the climate by exploiting recent advances in machine learning to enable improved projections and gain new insights from observations.

Project Highlights

Global temperature map


A benchmark dataset for the emulation of full-complexity climate models.

Learn more


Detecting ship tracks

Using machine learning to automatically detect the brightening effect that shipping can have on clouds.

Learn more

Emulator schematic

Model Emulation for Calibration

Developing climate model emulators for better parameter estimation and calibration.

Learn more


  • 24/03 - “Multi-Fidelity Residual Neural Processes for Scalable Surrogate Modeling” has been accepted to ICML 2024! It introduces a novel multi-fidelity neural process and apply it to climate projections - generating a temperature projection of ERA5 out to 2100. The paper is available on arXiv.
  • 24/03 - Prof. Watson-Parris gave a talk at the 2024 American Association for the Advancement of Science (AAAS) Annual Meeting in Denver on the use of Generative AI for Climate Science. It was covered by ACM.
  • 23/12 - The Climate Analytics Lab had a great time at AGU 2023! We gave two invited talks, presented a poster on AerChemMIP2 and had a great lunch with Andrew Ng to discuss AI for climate. We’re looking forward to AGU 2024!
  • 23/10 - The Climate Analytics Lab held our inaugral group meeting today - with cupcakes!
  • 23/08 - Prof. Watson-Parris is quoted in a Science article describing the climate impact of recent changes in shipping fuel regulations. Read more about the project here.
Show more