In the figure is the combined predictions of what the Earth’s climate will look like based on the climate models present in CMIP6 (Coupled Model Intercomparison Project Phase 6). They gathered the results of 134 different climate models, each of which were built upon different assumptions, but all of whom use a large amount of computational resources to attempt to model the behaviour of Earth’s climate in response to greenhouse gas emissions. The goal of these Earth System Models is to attempt to give as much information about the potential future of Earth’s climate to policy makers in order for us as a collective to be able to make sustainable choices to prevent drastic climate changes. But with so many different possible scenarios, using these ESMs to model every single socioeconomic pathway is too computationally expensive and time consuming. With these three deep learning models, we hope to create climate emulators that are both accurate and inexpensive to run.

Map of NorESM Model predictions