Steering Diffusion Priors with Sparse Observations for High-Resolution Temperature Downscaling
Abstract
Local heatwave hazard depends on fine-scale air temperature, but ground stations are sparse and reanalysis products such as ERA5 cannot resolve the terrain and land-surface contrasts that shape real heat exposure. We present a conditional diffusion emulator for high-resolution 2-m temperature downscaling, conditioned on static geography, a training climatology, exact-time ERA5 temperature, and solar and temporal features, guided at inference by score-based data assimilation (SDA): a differentiable Gaussian observation likelihood steers the diffusion score toward sparse revealed temperature observations without any retraining. On a controlled 32-case synthetic-grid protocol over AORC, guidance improves hidden-cell reconstruction over both ERA5 and a strong observation-proximal nearest-neighbor baseline once observation density reaches 1\% (RMSE 0.318 vs.\ 0.431~K, winning all 32 cases), while sparser regimes still favor direct interpolation. We further map the full guidance-strength landscape across three observation densities, showing that the optimal strength shifts systematically with density and that over-guiding causes sharp, predictable degradation -- giving a concrete operating recipe rather than a single untuned setting. The resulting fields are intended as a temperature layer for downstream heatwave-hazard products such as threshold exceedance and cumulative heat-burden. The present evidence is a controlled synthetic-grid validation; station-network and held-out-year evaluations are the next steps toward deployment.
Related
- Downscaling land surface temperature to high resolution using Sentinel-2 and Landsat in Google Earth Engine: a case study in Da Nang City
- Spatio-temporal precipitation downscaling at hourly resolution based on km-scale regional simulations
- Evaluating Cross-region Generalization for Wavelet-Diffusion Precipitation Downscaling
- Sea surface salinity downscaling using deep generative diffusion models
- Recurrent High-Resolution Satellite Observations Quantify Facility-Resolved Annual Large-Emitter Methane Emissions in the Permian Basin
BibTeX
@misc{avireddy2026steering,
title = {{Steering Diffusion Priors with Sparse Observations for High-Resolution Temperature Downscaling}},
author = {Anirudh Avireddy and Manmeet Singh and Shivanshi Singh and Ayush Raj and Saptarishi Dhanuka and Parthasarathi Mukhopadhyay and Sandeep Juneja},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.09247},
eprint = {2609.09247},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.09247},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Steering Diffusion Priors with Sparse Observations for High-Resolution Temperature Downscaling
AU - Anirudh Avireddy
AU - Manmeet Singh
AU - Shivanshi Singh
AU - Ayush Raj
AU - Saptarishi Dhanuka
AU - Parthasarathi Mukhopadhyay
AU - Sandeep Juneja
PY - 2026
DA - 2026/09/08
PB - arXiv
DO - 10.48550/arxiv.2609.09247
UR - https://arxiv.org/abs/2609.09247
AB - Local heatwave hazard depends on fine-scale air temperature, but ground stations are sparse and reanalysis products such as ERA5 cannot resolve the terrain and land-surface contrasts that shape real heat exposure. We present a conditional diffusion emulator for high-resolution 2-m temperature downscaling, conditioned on static geography, a training climatology, exact-time ERA5 temperature, and solar and temporal features, guided at inference by score-based data assimilation (SDA): a differentiable Gaussian observation likelihood steers the diffusion score toward sparse revealed temperature observations without any retraining. On a controlled 32-case synthetic-grid protocol over AORC, guidance improves hidden-cell reconstruction over both ERA5 and a strong observation-proximal nearest-neighbor baseline once observation density reaches 1\% (RMSE 0.318 vs.\ 0.431~K, winning all 32 cases), while sparser regimes still favor direct interpolation. We further map the full guidance-strength landscape across three observation densities, showing that the optimal strength shifts systematically with density and that over-guiding causes sharp, predictable degradation -- giving a concrete operating recipe rather than a single untuned setting. The resulting fields are intended as a temperature layer for downstream heatwave-hazard products such as threshold exceedance and cumulative heat-burden. The present evidence is a controlled synthetic-grid validation; station-network and held-out-year evaluations are the next steps toward deployment.
ER -