Residual-Guided Multi-Resolution Refinement of Foundation Models: A Case Study in Drought Forecasting

Wentao Gao, Jiuyong Li, Lin Liu, Thuc Duy Le, Jixue Liu, Yanchang Zhao, Yun Chen
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:33317-33339, 2026.

Abstract

Regional climate prediction presents unique challenges for time series foundation models, which typically process temporal patterns through single-pass inference. Expert climatologists, in contrast, employ multi-scale temporal analysis and iterative refinement based on systematic error diagnosis. We present RGMR (Residual-Guided Multi-Resolution Refinement), an inference-time framework that adapts pre-trained foundation models to perform structured coarse-to-fine refinement for climate forecasting without updating backbone parameters. Applied to drought forecasting using the Standardized Precipitation Evapotranspiration Index (SPEI), RGMR is architecture-agnostic across the three TSFM backbones evaluated per site (TimesFM, TimeGPT, TabPFN) and consistently lowers test-set MSE on three South Australian sites and three additional regions outside South Australia. Applied to TimesFM, the wrapper reduces one-month-ahead SPEI MSE by up to 18.9% across the three South Australian sites (mean reduction $\approx$18.7%). Overall, RGMR provides a practical route for deploying frozen TSFMs in regional climate forecasting workflows.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-gao26p, title = {Residual-Guided Multi-Resolution Refinement of Foundation Models: A Case Study in Drought Forecasting}, author = {Gao, Wentao and Li, Jiuyong and Liu, Lin and Le, Thuc Duy and Liu, Jixue and Zhao, Yanchang and Chen, Yun}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {33317--33339}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/gao26p/gao26p.pdf}, url = {https://proceedings.mlr.press/v306/gao26p.html}, abstract = {Regional climate prediction presents unique challenges for time series foundation models, which typically process temporal patterns through single-pass inference. Expert climatologists, in contrast, employ multi-scale temporal analysis and iterative refinement based on systematic error diagnosis. We present RGMR (Residual-Guided Multi-Resolution Refinement), an inference-time framework that adapts pre-trained foundation models to perform structured coarse-to-fine refinement for climate forecasting without updating backbone parameters. Applied to drought forecasting using the Standardized Precipitation Evapotranspiration Index (SPEI), RGMR is architecture-agnostic across the three TSFM backbones evaluated per site (TimesFM, TimeGPT, TabPFN) and consistently lowers test-set MSE on three South Australian sites and three additional regions outside South Australia. Applied to TimesFM, the wrapper reduces one-month-ahead SPEI MSE by up to 18.9% across the three South Australian sites (mean reduction $\approx$18.7%). Overall, RGMR provides a practical route for deploying frozen TSFMs in regional climate forecasting workflows.} }
Endnote
%0 Conference Paper %T Residual-Guided Multi-Resolution Refinement of Foundation Models: A Case Study in Drought Forecasting %A Wentao Gao %A Jiuyong Li %A Lin Liu %A Thuc Duy Le %A Jixue Liu %A Yanchang Zhao %A Yun Chen %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-gao26p %I PMLR %P 33317--33339 %U https://proceedings.mlr.press/v306/gao26p.html %V 306 %X Regional climate prediction presents unique challenges for time series foundation models, which typically process temporal patterns through single-pass inference. Expert climatologists, in contrast, employ multi-scale temporal analysis and iterative refinement based on systematic error diagnosis. We present RGMR (Residual-Guided Multi-Resolution Refinement), an inference-time framework that adapts pre-trained foundation models to perform structured coarse-to-fine refinement for climate forecasting without updating backbone parameters. Applied to drought forecasting using the Standardized Precipitation Evapotranspiration Index (SPEI), RGMR is architecture-agnostic across the three TSFM backbones evaluated per site (TimesFM, TimeGPT, TabPFN) and consistently lowers test-set MSE on three South Australian sites and three additional regions outside South Australia. Applied to TimesFM, the wrapper reduces one-month-ahead SPEI MSE by up to 18.9% across the three South Australian sites (mean reduction $\approx$18.7%). Overall, RGMR provides a practical route for deploying frozen TSFMs in regional climate forecasting workflows.
APA
Gao, W., Li, J., Liu, L., Le, T.D., Liu, J., Zhao, Y. & Chen, Y.. (2026). Residual-Guided Multi-Resolution Refinement of Foundation Models: A Case Study in Drought Forecasting. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:33317-33339 Available from https://proceedings.mlr.press/v306/gao26p.html.

Related Material