mlsanityData to 7 Oct 2026

Beyond Pointwise Error: A Multi-Metric Evaluation of Spatial Climate Downscaling

arXiv 2610.01579 ↗First posted 1 Oct 2026cs.LG cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Climate downscaling aims to reconstruct fine scale spatial fields from coarse resolution inputs. Evaluating the quality of these reconstructions is challenging: low pointwise error can come at the cost of fine scale variability, while realistic spatial variability can be achieved with inaccurate local structures. The evaluation metric can therefore change which method appears to perform best. This work presents a multi metric benchmark comparing five spatial downscaling methods on ERA5 temperature, wind, and precipitation fields. Five criteria assess complementary properties: pointwise error, structural similarity, distribution error, spectral error, and gradient error. The results reveal a systematic trade off between spatial fidelity and fine scale variability. Some methods perform best on pointwise and spatially aligned metrics, but lose high frequency content, while others preserve substantially more spectral variability at the cost of less accurately positioned local structures. Consequently, method rankings change across metrics and variables. These results show that there is no single best downscaling method. Multi metric evaluation is therefore essential for assessing which properties of a climate field are preserved.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 17 of its 29 references that are in the map. First posted on arXiv 1 Oct 2026.

Builds on

17 of 29 references.

References · 17

  1. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  2. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 11,979
  3. 2018Representation Learning with Contrastive Predictive Codingcited by 7,662
  4. 2013Sinkhorn Distances: Lightspeed Computation of Optimal Transportation Distancescited by 2,277
  5. 2020Implicit Neural Representations with Periodic Activation Functionscited by 2,214
  6. 2016Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Networkcited by 1,898
  7. 2014Image Super-Resolution Using Deep Convolutional Networkscited by 1,218
  8. 2015Accurate Image Super-Resolution Using Very Deep Convolutional Networkscited by 902
  9. 2018Computational Optimal Transportcited by 765
  10. 2017The Perception-Distortion Tradeoffcited by 582
  11. 2017Road Extraction by Deep Residual U-Netcited by 333
  12. 2013Gradient Magnitude Similarity Deviation: A Highly Efficient Perceptual Image Quality Indexcited by 193
5 more
  1. 2023Operator Learning with Neural Fields: Tackling PDEs on General Geometriescited by 80
  2. 2022A Generative Deep Learning Approach to Stochastic Downscaling of Precipitation Forecastscited by 64
  3. 2021A Theory of the Distortion-Perception Tradeoff in Wasserstein Spacecited by 52
  4. 2023Time Series Continuous Modeling for Imputation and Forecasting with Implicit Neural Representationscited by 16
  5. 2025Frequency-Aware Vision Transformers for High-Fidelity Super-Resolution of Earth System Modelscited by 1

Built on by

Not cited yet.