In collaboration with the National Institute for Environmental Studies, we have developed a downscaling method that uses machine learning to refine climate prediction information. This method provides detailed information with a spatial resolution approximately 50 times finer than that of coarse global climate predictions. The research team verified the method's ability to rapidly and precisely predict local temperature and precipitation statistics, as well as the relations between climate phenomena in distant locations, using real data. This research is expected to contribute to the assessment of climate change impacts while accounting for spatial spread. For the full text, please refer to the link above.

 

The findings of this study were published in the Scientific Reports, a multidisciplinary journal from the Nature Publishing Group, in its online edition.

 

 Title: Deep generative model super-resolves spatially correlated multiregional climate data
 Authors: Norihiro Oyama, Noriko N. Ishizaki, Satoshi Koide, Hiroaki Yoshida
 Journal Name: Scientific Reports
 Published: April 25, 2023
 https://doi.org/10.1038/s41598-023-32947-0