Deep learning super-resolution for temperature data downscaling: a comprehensive study using residual networks
Frontiers in Climate · 13.05.2025 · 7
DOI: 10.3389/fclim.2025.1572428 ↗Kurz gefasst
Fachartikel in der Zeitschrift Frontiers in Climate, online erschienen am 13.05.2025. Verfasst von Shailesh Kumar Jha und 4 weiteren Autor:innen. Die Arbeit wurde bisher 21-mal zitiert (Crossref) und nennt 79 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Extreme weather events such as heatwaves, cyclones, floods, wildfires, and droughts are becoming more frequent due to climate change. Climate change causes shifts in biodiversity and impacts agriculture, forest ecosystems, and water resources at a regional scale. However, to study those impacts at the regional scale, the spatial resolution provided by the general circulation models (GCMs) and reanalysis products is inadequate. This study evaluates advanced deep learning models for downscaling European Center for Medium-Range Weather Forecasts (ECMWF) Reanalysis v5 (ERA5) 2-m temperature data by a factor of 10 (i.e., ranging approximately from 250 to 25 km resolution) for the region spanning 50° to 100° E and 0° to 50° N. We concentrate on gradually improving downscaling models with the help of residual networks. We compare the baseline Super-Resolution Convolutional Neural Network (SRCNN) model with two advanced models: Very Deep Super-Resolution (VDSR) and Enhanced Deep Super-Resolution (EDSR) to assess the impact of residual networks and architectural improvements. The results indicate that VDSR and EDSR significantly outperform SRCNN. Specifically, VDSR increases the Peak Signal-to-Noise Ratio (PSNR) by 4.27 dB and EDSR by 5.23 dB. These models also enhance the Structural Similarity Index Measure (SSIM) by 0.1263 and 0.1163, respectively, indicating better image quality. Furthermore, improvements in the 3°C error threshold are observed, with VDSR and EDSR showing increases of 2.10 and 2.16%, respectively. An explainable artificial intelligence (AI) technique called saliency map analysis provided insights into model performance. Complex terrain areas, such as the Himalayas and the Tibetan Plateau, benefit the most from these advancements. These findings suggest that advanced deep learning models employing residual networks, such as VDSR and EDSR, significantly enhance temperature data accuracy over SRCNN. This approach holds promise for future applications in downscaling other atmospheric variables.
Zitieren
APA
Jha, S. K., Gupta, V., Sharma, P. J., Mishra, A., & Joshi, S. (2025). Deep learning super-resolution for temperature data downscaling: a comprehensive study using residual networks. Frontiers in Climate, 7. https://doi.org/10.3389/fclim.2025.1572428
Vancouver
Jha SK, Gupta V, Sharma PJ, Mishra A, Joshi S. Deep learning super-resolution for temperature data downscaling: a comprehensive study using residual networks. Frontiers in Climate. 2025;7. doi:10.3389/fclim.2025.1572428
BibTeX
@article{Jha2025,
author = {Jha, Shailesh Kumar and Gupta, Vivek and Sharma, Priyank J. and Mishra, Anurag and Joshi, Saksham},
title = {Deep learning super-resolution for temperature data downscaling: a comprehensive study using residual networks},
journal = {Frontiers in Climate},
year = {2025},
volume = {7},
doi = {10.3389/fclim.2025.1572428}
}
EUVIMED