Warping Earth Observations for better ice labeling in the Marginal Marginal Ice Zone

2026-08-12

AI Summary

This study addresses the challenge of pixel-level misalignment between multi-source satellite images in dynamic marginal ice zones, which hinders effective multimodal fusion and accurate sea ice segmentation. To overcome this, the authors propose a novel mutual information-based warping framework—the first to apply this approach for spatial alignment between Sentinel-1 synthetic aperture radar and MODIS visible/thermal infrared imagery. Integrated with sparse point annotations, the method enables high-precision dense semantic segmentation, departing from conventional training paradigms that rely on coarse-grained ice charts. Evaluated on a dataset comprising 43 scenes with 2,088 pixel-level labels derived from 7,046 expert-annotated points, experiments demonstrate significantly improved segmentation accuracy after alignment, effectively achieving robust generalization from sparse annotations to dense predictions.

Abstract

Multimodal satellite imagery provides complementary information for Earth Observation, but accurately combining heterogeneous sensors remains challenging in dynamic environments. Fast-changing regions, such as the Antarctic marginal ice zone, cannot fully exploit multimodal information from different satellite sensors because surface features move between image acquisitions. This spatial and temporal mismatch challenges effective perceptual grounding, violating the assumption of pixel-level correspondence that underpins most multimodal reasoning and downstream classification pipelines. Antarctic sea ice provides a challenging benchmark due to the rapid, heterogeneous drift of individual ice floes and the differing responses of sea ice to radar, visible and thermal sensing modalities. Accurate, dense supervision of sea ice remains scarce because generating pixel-wise labels requires time-consuming expert interpretation of noisy data, leading to historical reliance on coarse-resolution maritime ice charts for model training. This paper presents a novel architecture based on mutual information warping to align multi-satellite (Sentinel-1 and MODIS platforms) multimodal (visible, thermal, radar) satellite scenes. To demonstrate the approach, we introduce a sparse expert-labeled dataset of 2,088 pixel-wise annotations (7,046 expert point classifications) located at the ice-water margin interface across 43 scenes. Our results demonstrate that spatially grounding and aligning modalities prior to segmentation improves classification accuracy, and enables accurate, dense sea ice segmentation from sparse point-wise supervision.

Problem

• multimodal satellite imagery
• spatial-temporal misalignment
• marginal ice zone
• pixel-wise labeling
• sea ice classification

Innovation

• mutual information warping
• multimodal alignment
• marginal ice zone
• sparse supervision
• sea ice segmentation