> Home > Projects > Embedding machine-learnt sub-grid variability improves climate model precipitation patterns

Embedding machine-learnt sub-grid variability improves climate model precipitation patterns

Using Gaussian processes to model sub-grid variability induced by clouds and injecting this stochastically at run-time into climate simulations.

September 1, 2024

Upcoming talk on Friday 7th March at SIAM CSE25 in Fort Worth, Texas.

See publications for this paper. Published by Nature Communications Earth & Environment.

Project code available on GitHub.

See /talks/ATI-CI24 for a recorded presentation of this work from at Climate Informatics 2024. Also previously presented at SIAM UQ24 in Trieste, Italy and at RMetS Annual Weather and Climate conference 2024 in Reading, UK.