Recent paper in GRL from Elizabeth Barnes et al: Viewing forced climate patterns through an AI Lens
A bit off topic maybe because 'pure' research, but interesting nonetheless in that it shows how AI / ML is becoming increasingly adopted in climate science
From the abstract:
Many problems in climate science require the identification of signals amidst a sea of climate “noise” and across a variety of models which can disagree with one another. Here, we demonstrate that machine learning techniques, specifically artificial neural networks, can help identify forced patterns of temperature and precipitation within climate model simulations as well as the observations. In fact, the neural network is able to identify patterns of forced change of surface temperature as early as the 1960’s in climate model simulations. The results shown here are strongly suggestive of the potential power of machine learning for climate research.
@NicolasF Interesting. The key word that it always comes back to for me with AI is "prediction". If there is a task that can be described as prediction, it is likely that it can be learned by a model. Lots of tasks under the domain of climate solutions require predictive capabilities, including your own which you highlighted in the other thread
@Rob-Bennett yes ! they actually set up the task as a prediction / classification task in this study, where the target / label to predict is the year where the patterns are coming from, very clever way to look at the issue of separating the climate change signal from the background, climate variability 'noise' ...