BCM4RCM: Bayesian Committee Machines for Regional Climate Models
An ensembling method to combine different regional climate model outputs and produce principled uncertainty estimates of precipitation under different climate scenarios
Featured projects
I’ve been lucky enough to work with some amazing people on a wide variety of fascinating topics.
Predicting future precipitation in the Upper Indus Basin using Gaussian Processes
Large-scale atmospheric variables are used to make 15-year projections where flexible non-stationary techniques are contrasted with methods incorporating domain knowledge
Beyond intuition, a framework for applying Gaussian Processes to real-word data
Formalising the decision-making process of experienced Gaussian Processes users with an emphasis on kernel design and computational scalability
Pyrocast: A machine learning framework for forcasting pyroCb clouds
A pipeline for the identification, forecasting and causal prediction of pyrocumulonimbus clouds generated by extreme wildfires
Narrowing precipitation uncertainty over High Mountain Asia
Downscaling precipitation using Multi-Fidelity Gaussian Processes by combining data from multiple sources to increase prediction accuracy and provide principled uncertainty distribution
Cambridgeshire Decarbonisation Fund: Net-Zero by 2050
White paper on investment opportunities to support locally in community infrastructure and nature-based projects that reduce carbon emissions at their source or actively sequester carbon
cloud-id: Cloud identification over polar regions
Deep learning model to identify clouds over polar regions using satellite data from the Sentinel 3 SLSTR instrument