Close

Presentation

Machine Learning Day
:
Detecting anthropogenic cloud perturbations with deep learning
Event Type
Machine Learning Day
Passes
TimeWednesday, June 24th2:45pm - 3:15pm
LocationPanorama 3
DescriptionTwo of the most pressing climate questions we currently face are that of the effect of anthropogenic aerosol on the climate system, and the feedbacks which clouds exert on the changing climate. The large uncertainties in both effects hinder our efforts to accurately predict future climate warming.

Aerosols provide the `seeds' on which cloud droplets form, and changes in the amount of aerosol available to a cloud can change its brightness and other physical properties such as optical thickness and spatial extent. Large uncertainties in this effect persist because of the difficulty in measuring the very small changes in cloud properties due to aerosols as compared to the main, meteorological, drivers of cloud formation. Clouds play a large role in moderating global temperatures and so even these small perturbations can lead to significant amounts of cooling or warming over long timescales.

Here we will describe recent work using deep convolutional neural networks to look for two particular perturbations in clouds due to anthropogenic aerosol and assess their properties and prevalence, providing valuable insights into their climatic effects. We will also discuss a semi-supervised approach to cloud classification combining existing, complementary satellite observations and providing an improved understanding of the cloud-climate system. New machine learning tools could allow us to unpick these relationships, determine the role of aerosol in driving cloud properties and hence improve our understanding of the changing climate.