Sea surface salinity downscaling using deep generative diffusion models
Abstract
High-resolution satellite observations are essential for studying fine-scale ocean processes. Yet, present satellite sea surface salinity (SSS) products remain too coarse to resolve many fine-scale structures. We investigate denoising diffusion models as a generative framework for SSS downscaling in a controlled proof-of-concept experiment based on GLORYS reanalysis fields. A multichannel diffusion prior is trained on 1/12° SSS, sea surface temperature (SST), and sea surface height (SSH) fields, and is then conditioned at inference time on a synthetically degraded coarse SSS observation (1/3°) together with high-resolution (1/12°) auxiliary SST and/or SSH. Conditioning is performed through pseudo-inverse guidance, which steers the generated samples toward states that are compatible with the coarse observation while remaining within the learned GLORYS-consistent multivariate distribution. We also test a gradient-enhancement procedure designed to increase contrast during inference. Experiments in the Gulf Stream region compare models conditioned on SST only, SSH only, and both variables. Validation over the year 2020 uses root-mean-square error (RMSE), structural similarity (SSIM), gradient distributions, and temporal Fourier spectra. In the present GLORYS configuration, conditioning on SST substantially improves accuracy relative to SSH alone; combining SST and SSH yields further gains, comparable to a strong convolutional baseline under RMSE/SSIM, while additionally providing an ensemble of plausible reconstructions. The gradient-enhanced sampler increases structural contrast but can risk amplifying part of the variability, illustrating a trade-off between pixel-wise accuracy and structural realism. Overall, the results support guided diffusion as a promising framework for SSS downscaling and uncertainty-aware reconstruction, while showing that transfer to real satellite SSS products will require product-aware observation operators, uncertainty weighting, and independent in-situ validation.
Related
- Kilometer-Scale AI Downscaling of Atlantic Hurricanes with Generative Ensembles
- Stress-Testing Dynamical and Generative Downscaling Using Subseasonal Extreme Precipitation Forecasts
- Deep learning-based downscaling of PM2.5, NO2 and O3 over Europe across emission-reduction scenarios
- Generative modelling for multivariate spatio-temporal downscaling of monthly climate fields
- EnScale: Temporally‐Consistent Multivariate Generative Downscaling via Proper Scoring Rules
BibTeX
@article{forestier2026surface,
title = {{Sea surface salinity downscaling using deep generative diffusion models}},
author = {Enzo Forestier and Luther Ollier and Roy El Hourany and Jacqueline Boutin and Carlos Mejia and Sylvie Thiria},
journal = {Ocean Science},
year = {2026},
month = aug,
doi = {10.5194/os-22-2357-2026},
url = {https://hal.science/hal-05820459v1},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Sea surface salinity downscaling using deep generative diffusion models
AU - Enzo Forestier
AU - Luther Ollier
AU - Roy El Hourany
AU - Jacqueline Boutin
AU - Carlos Mejia
AU - Sylvie Thiria
PY - 2026
DA - 2026/08/05
JO - Ocean Science
DO - 10.5194/os-22-2357-2026
UR - https://hal.science/hal-05820459v1
AB - High-resolution satellite observations are essential for studying fine-scale ocean processes. Yet, present satellite sea surface salinity (SSS) products remain too coarse to resolve many fine-scale structures. We investigate denoising diffusion models as a generative framework for SSS downscaling in a controlled proof-of-concept experiment based on GLORYS reanalysis fields. A multichannel diffusion prior is trained on 1/12° SSS, sea surface temperature (SST), and sea surface height (SSH) fields, and is then conditioned at inference time on a synthetically degraded coarse SSS observation (1/3°) together with high-resolution (1/12°) auxiliary SST and/or SSH. Conditioning is performed through pseudo-inverse guidance, which steers the generated samples toward states that are compatible with the coarse observation while remaining within the learned GLORYS-consistent multivariate distribution. We also test a gradient-enhancement procedure designed to increase contrast during inference. Experiments in the Gulf Stream region compare models conditioned on SST only, SSH only, and both variables. Validation over the year 2020 uses root-mean-square error (RMSE), structural similarity (SSIM), gradient distributions, and temporal Fourier spectra. In the present GLORYS configuration, conditioning on SST substantially improves accuracy relative to SSH alone; combining SST and SSH yields further gains, comparable to a strong convolutional baseline under RMSE/SSIM, while additionally providing an ensemble of plausible reconstructions. The gradient-enhanced sampler increases structural contrast but can risk amplifying part of the variability, illustrating a trade-off between pixel-wise accuracy and structural realism. Overall, the results support guided diffusion as a promising framework for SSS downscaling and uncertainty-aware reconstruction, while showing that transfer to real satellite SSS products will require product-aware observation operators, uncertainty weighting, and independent in-situ validation.
SN - 1812-0784
SN - 1812-0792
ER -