The Devil is in the Details: Spatial and Temporal Super-Resolution of Global Climate Models using Deep Learning
The University of Michigan team ("Bayes and Blue") repository for ProjectX 2020, an undergraduate research competition focused on the use of machine learning in climate modeling.
Physics-based global climate simulations are computationally expensive and limited to low spatial and temporal resolutions, making it difficult to predict and track highly localized extreme weather phenomena. To overcome these limitations, we present a novel application of super-resolution using deep learning to increase the resolution of global climate models in both space and time. In this project, we demonstrate the potential to reduce climate simulation computation and storage requirements by two orders of magnitude, as well as democratize relevant and actionable climate information for disaster responses.
We used a subset of the ExtremeWeather dataset, consisting of images produced from the CAM5 (Community Atmospheric Model v5) climate simulation. We also used the NCEP dataset, consisting of climate maps generated by a combination of real-world observations and numerical weather prediction model output from 1948 to present.
The exact subset of data we used for the project can be found at here.
Our research paper, "The Devil is in the Details: Spatial and Temporal Super-Resolution of Global Climate Models using Deep Learning", can be found here.
Eric Chen, Sanjeev Raja
Yue (Amanda) Yao, Zhizhuo Zhou
Ziwei Tian, Anh Tuan Tran
Dr. Sindhu Kutty, CSE Department