Geographical Context Matters: Bridging Fine and Coarse Spatial Information to Enhance Continental Land Cover Mapping

2025-04-16

AI Summary

To address poor cross-regional generalization in remote sensing land use/cover classification caused by insufficient geographic context modeling, this paper proposes a two-level geographic representation disentanglement framework. During training, it jointly models fine-grained geographic coordinates (latitude/longitude) and coarse-grained biogeographic zone priors; at inference, only latitude/longitude inputs are required. We design a geographic-information-guided feature disentanglement strategy and a dual-path spatial embedding scheme, first revealing geometric consistency between the model’s latent space and real-world geographic structures. Employing a lightweight MLP architecture and a leave-one-region-out evaluation protocol, our method achieves an 8.2% improvement in cross-regional extrapolation accuracy on open-field remote sensing datasets, with significantly enhanced robustness in data-scarce regions. This validates the critical value of explicit geographic context modeling for large-scale mapping.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search

Abstract

Land use and land cover mapping from Earth Observation (EO) data is a critical tool for sustainable land and resource management. While advanced machine learning and deep learning algorithms excel at analyzing EO imagery data, they often overlook crucial geospatial metadata information that could enhance scalability and accuracy across regional, continental, and global scales. To address this limitation, we propose BRIDGE-LC (Bi-level Representation Integration for Disentangled GEospatial Land Cover), a novel deep learning framework that integrates multi-scale geospatial information into the land cover classification process. By simultaneously leveraging fine-grained (latitude/longitude) and coarse-grained (biogeographical region) spatial information, our lightweight multi-layer perceptron architecture learns from both during training but only requires fine-grained information for inference, allowing it to disentangle region-specific from region-agnostic land cover features while maintaining computational efficiency. To assess the quality of our framework, we use an open-access in-situ dataset and adopt several competing classification approaches commonly considered for large-scale land cover mapping. We evaluated all approaches through two scenarios: an extrapolation scenario in which training data encompasses samples from all biogeographical regions, and a leave-one-region-out scenario where one region is excluded from training. We also explore the spatial representation learned by our model, highlighting a connection between its internal manifold and the geographical information used during training. Our results demonstrate that integrating geospatial information improves land cover mapping performance, with the most substantial gains achieved by jointly leveraging both fine- and coarse-grained spatial information.

Problem

• Enhancing land cover mapping by integrating multi-scale geospatial data
• Disentangling region-specific and region-agnostic features for better accuracy
• Improving scalability and performance in continental land cover classification

Innovation

• Integrates multi-scale geospatial information
• Uses fine and coarse spatial data
• Lightweight MLP architecture for efficiency