Lightweight Wrappers for Adapting Time Series Foundation Models to Regional Drought Forecasting

2026-07-19

AI Summary

This study addresses the challenges of regional drought forecasting under constraints such as inaccessible weights of general-purpose time series foundation models, scarce local data, and limited computational resources, which render conventional fine-tuning impractical. To overcome these limitations, the authors propose a lightweight, black-box adaptation framework that operates without access to model parameters or fine-tuning. During inference, the framework dynamically adapts frozen foundation models through two plug-and-play modules: SMR² (Stationarity-aware Multi-Resolution Residual), which performs multi-scale decomposition and adaptive ensemble learning, and MBB (Moving Block Bootstrap), which preserves temporal dependencies while generating consistent perturbed residuals. Evaluated on one-month-ahead SPEI prediction across multiple locations in South Australia, the approach reduces the average mean squared error of several frozen backbone models by up to 26%, substantially improving predictive accuracy while remaining suitable for resource-constrained settings.

Abstract

Large \emph{Time Series Foundation Models} (TSFMs) demonstrate strong zero-shot forecasting capabilities across diverse domains. However, their application to regional climate forecasting faces practical challenges: model weights are often proprietary, local training records are limited, and computational budgets are constrained, making traditional fine-tuning approaches infeasible. To address these constraints, we introduce a lightweight, black-box adaptation framework (requiring no access to backbone parameters and no backbone fine-tuning) that enhances frozen TSFMs at inference time through two plug-and-play wrappers: \textbf{SMR\textsuperscript{2}} (Stationarity aware multi-resolution Residual), which decomposes the input into multi-resolution temporal views, learns stride specific residual corrections that capture regional dynamics, then adaptively ensembles them into a single forecast, and \textbf{MBB} (Moving Block Bootstrap), which preserves temporal dependencies through block resampling and ensembles over temporally coherent residual perturbations to stabilize the point forecast. Both wrappers instantiate the same bagging style principle: they build diverse views of the input or its residuals, forecast each with the same frozen backbone, and aggregate, so all adaptation comes from inference time ensembling rather than any weight update. Evaluated on one month ahead Standardized Precipitation Evapotranspiration Index (SPEI) prediction across multiple sites in South Australia, our framework consistently improves forecasting performance across several backbone models, demonstrating up to 26\% mean squared error (MSE) reduction over the corresponding frozen backbone while enabling practical deployment in resource constrained regional forecasting systems.

Problem

• Time Series Foundation Models
• regional drought forecasting
• black-box adaptation
• limited training data
• computational constraints

Innovation

• Time Series Foundation Models
• Lightweight Adaptation
• Inference-time Ensembling
• Regional Drought Forecasting
• Black-box Adaptation