Uncertainty-Aware Test-Time Adaptation for Cross-Region Spatio-Temporal Fusion of Land Surface Temperature

2026-04-05

AI Summary

This study addresses the degradation in generalization performance of deep learning models for spatiotemporal land surface temperature fusion across regions, primarily caused by domain shift—a challenge exacerbated by the lack of effective test-time adaptation methods for regression tasks in remote sensing. To this end, this work introduces test-time adaptation to remote sensing regression for the first time and proposes a lightweight, source-data- and target-label-free adaptive framework. By fine-tuning only the fusion module of a pretrained model and integrating cognitive uncertainty guidance, land use/cover consistency constraints, and a bias correction mechanism, the method achieves significant improvements across four climatically diverse regions—Rome, Cairo, Madrid, and Montpellier—reducing average RMSE and MAE by 24.2% and 27.9%, respectively, within just 10 adaptation iterations.

Abstract

Deep learning models have shown great promise in diverse remote sensing applications. However, they often struggle to generalize across geographic regions unseen during training due to domain shifts. Domain shifts occur when data distributions differ between the training region and new target regions, due to variations in land cover, climate, and environmental conditions. Test-time adaptation (TTA) has emerged as a solution to such shifts, but existing methods are primarily designed for classification and are not directly applicable to regression tasks. In this work, we address the regression task of spatio-temporal fusion (STF) for land surface temperature estimation. We propose an uncertainty-aware TTA framework that updates only the fusion module of a pre-trained STF model, guided by epistemic uncertainty, land use and land cover consistency, and bias correction, without requiring source data or labeled target samples. Experiments on four target regions with diverse climates, namely Rome in Italy, Cairo in Egypt, Madrid in Spain, and Montpellier in France, show consistent improvements in RMSE and MAE for a pre-trained model in Orléans, France. The average gains are 24.2% and 27.9%, respectively, even with limited unlabeled target data and only 10 TTA epochs.

Problem

• domain shift
• spatio-temporal fusion
• land surface temperature
• test-time adaptation
• regression

Innovation

• Test-Time Adaptation
• Uncertainty-Aware
• Spatio-Temporal Fusion
• Land Surface Temperature
• Domain Shift