mlsanityData to 14 Oct 2026

Downscaling climate projections to 1 km with single-image super resolution

arXiv 2509.21399 ↗First posted 24 Sep 2025cs.CV cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

High-resolution climate projections are essential for local decision-making. However, available climate projections have low spatial resolution (e.g. 12.5 km), which limits their usability. We address this limitation by leveraging single-image super-resolution models to statistically downscale climate projections to 1-km resolution. Since high-resolution climate projections are unavailable, we train models on a high-resolution observational gridded data set and apply them to low-resolution climate projections. We cannot evaluate downscaled climate projections with common metrics (e.g. pixel-wise root-mean-square error) because we lack ground-truth high-resolution climate projections. Therefore, we evaluate climate indicators computed at weather station locations. Experiments on daily mean temperature demonstrate that single-image super-resolution models can downscale climate projections without increasing the error of climate indicators compared to low-resolution climate projections.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 15 of its 26 references that are in the map. First posted on arXiv 24 Sep 2025.

Builds on

15 of 26 references.

References · 15

  1. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,139
  2. 2020Fourier Neural Operator for Parametric Partial Differential Equationscited by 2,908
  3. 2017Enhanced Deep Residual Networks for Single Image Super-Resolutioncited by 1,552
  4. 2014Image Super-Resolution Using Deep Convolutional Networkscited by 1,224
  5. 2023WeatherBench 2: A benchmark for the next generation of data-driven global weather modelscited by 236
  6. 2023Residual Corrective Diffusion Modeling for Km-scale Atmospheric Downscalingcited by 110
  7. 2017DeepSD: Generating High Resolution Climate Change Projections through Single Image Super-Resolutioncited by 82
  8. 2022Increasing the accuracy and resolution of precipitation forecasts using deep generative modelscited by 35
  9. 2023Fourier Neural Operators for Arbitrary Resolution Climate Data Downscalingcited by 26
  10. 2023SuperBench: A Super-Resolution Benchmark Dataset for Scientific Machine Learningcited by 18
  11. 2022Deep generative model super-resolves spatially correlated multiregional climate datacited by 9
  12. 2024Evaluating the transferability potential of deep learning models for climate downscalingcited by 8
3 more
  1. 2023Multi-variable Hard Physical Constraints for Climate Model Downscalingcited by 6
  2. 2024On the Effectiveness of Neural Operators at Zero-Shot Weather Downscalingcited by 5
  3. 2024Transformer based super-resolution downscaling for regional reanalysis: Full domain vs tiling approachescited by 4

Built on by

Not cited yet.