Authors: Guiye Li, Guofeng Cao

Published on: February 21, 2024

Impact Score: 7.6

Arxiv code: Arxiv:2402.14049

Summary

• What is new: A conditional GAN-based method for extreme downscaling of gridded climate datasets, capable of producing multiple high-resolution outcomes to assess model uncertainty.
• Why this is important: Existing climate datasets are at coarse resolutions, limiting their usefulness for detailed climate change research and analysis.
• What the research proposes: Using a conditional GAN-based method to downscale gridded climate datasets to higher resolutions while considering inherent uncertainties.
• Results: Successful downscaling of climate datasets (wind velocity and solar irradiance) with high accuracy and the ability to explore model uncertainties.

Technical Details

Technological frameworks used: Generative Adversarial Networks (GANs), Conditional GANs

Models used: ATP kriging, DIP, EDSR, ESRGAN, PhIRE GAN

Data used: Gridded climate datasets (wind velocity, solar irradiance)

Potential Impact

Climate research organizations, renewable energy companies, environmental policy makers, and urban planning agencies

Want to implement this idea in a business?

We have generated a startup concept here: ClimateRefine.