Time Series Classification of Supraglacial Lakes Evolution over Greenland Ice Sheet

2024-10-08

International Conference on Machine Learning and Applications

AI Summary

Identifying supraglacial lake evolution patterns on the Greenland Ice Sheet faces challenges of limited labeled samples and difficulty in modeling highly nonlinear time-series dynamics. Method: This paper proposes RPS-GMM, a lightweight time-series classification framework that combines Reconstruction Phase Space (RPS) for dimensionality reduction of high-dimensional remote sensing time series with Gaussian Mixture Modeling (GMM) for discriminating among three evolution patterns—refreeze, drain, and buried. It introduces a novel “one-sample-per-class” training paradigm, eliminating reliance on large-scale annotated datasets. Results: By fusing Sentinel-1 (microwave) and Sentinel-2 (optical) time-series data, RPS-GMM achieves 89.70% accuracy at the full ice-sheet scale—surpassing Sentinel-1 alone (85.46%) and outperforming mainstream machine learning and deep learning methods. These results demonstrate superior robustness and generalization capability under small-sample conditions.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Computer Vision: Remote Sensing / Geospatial AI
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

The Greenland Ice Sheet (GrIS) has emerged as a significant contributor to global sea level rise, primarily due to increased meltwater runoff. Supraglacial lakes, which form on the ice sheet surface during the summer months, can impact ice sheet dynamics and mass loss; thus, better understanding these lakes' seasonal evolution and dynamics is an important task. This study presents a computation-ally efficient time series classification approach that uses Gaussian Mixture Models (GMMs) of the Reconstructed Phase Spaces (RPSs) to identify supraglacial lakes based on their seasonal evolution: 1) those that refreeze at the end of the melt season, 2) those that drain during the melt season, and 3) those that become buried, remaining liquid insulated a few meters beneath the surface. Our approach uses time series data from the Sentinel-l and Sentinel-2 satellites, which utilize microwave and visible radiation, respectively. Evaluated on a GrIS-wide dataset, the RPS-GMM model, trained on a single representative sample per class, achieves 85.46% accuracy with Sentinel-l data alone and 89.70% with combined Sentinel-l and Sentinel-2 data. This performance significantly surpasses existing machine learning and deep learning models which require a large training data. The results demonstrate the robustness of the RPS-GMM model in capturing the complex temporal dynamics of supraglaciallakes with minimal training data.

Problem

• Classify seasonal evolution of supraglacial lakes on Greenland Ice Sheet.
• Improve understanding of supraglacial lakes' impact on ice sheet dynamics.
• Develop efficient time series classification with minimal training data.

Innovation

• Uses Gaussian Mixture Models for classification
• Leverages Sentinel-1 and Sentinel-2 satellite data
• Achieves high accuracy with minimal training data