Residual-Guided Multi-Resolution Refinement of Foundation Models: A Case Study in Drought Forecasting
2026-07-20 • Machine Learning
Machine Learning
AI summaryⓘ
The authors introduce RGMR, a method that helps existing climate prediction models improve their forecasts by refining predictions step-by-step without changing the original model. Instead of just making one guess, RGMR breaks down the problem into coarse and fine steps to better capture complex patterns over time. They tested RGMR on drought forecasting and found it consistently reduced errors across different models and locations. This approach makes it easier to use current models for regional climate predictions without retraining them.
regional climate predictiontime series foundation modelsinference-time refinementStandardized Precipitation Evapotranspiration Index (SPEI)drought forecastingmulti-scale temporal analysiscoarse-to-fine refinementmodel freezingmachine learning model adaptation
Authors
Wentao Gao, Jiuyong Li, Lin Liu, Thuc Duy Le, Jixue Liu, Yanchang Zhao, Yun Chen
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.