Localized, High-resolution Geographic Representations with Slepian Functions

2026-01-30

AI Summary

This work addresses the limitation of existing geospatial encoding methods, which uniformly distribute representational capacity across the globe and thus struggle to simultaneously capture high-resolution local details and global contextual information. To overcome this, the study introduces spherical Slepian functions into geographic encoding for the first time, proposing a locally focused Slepian encoder. Furthermore, a hybrid Slepian–spherical harmonic encoder is designed to effectively integrate local detail with global structure while preserving pole safety and fidelity to spherical distances. The proposed approach consistently outperforms current baselines across five diverse geospatial tasks—including classification, regression, and image enhancement—and demonstrates robust performance gains across multiple neural network architectures.

Technology Category

• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Computer Vision: Remote Sensing / Geospatial AI
• Search and Optimization: Local Search

Application Category

• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Security and Privacy: Privacy-enhancing technologies

Abstract

Geographic data is fundamentally local. Disease outbreaks cluster in population centers, ecological patterns emerge along coastlines, and economic activity concentrates within country borders. Machine learning models that encode geographic location, however, distribute representational capacity uniformly across the globe, struggling at the fine-grained resolutions that localized applications require. We propose a geographic location encoder built from spherical Slepian functions that concentrate representational capacity inside a region-of-interest and scale to high resolutions without extensive computational demands. For settings requiring global context, we present a hybrid Slepian-Spherical Harmonic encoder that efficiently bridges the tradeoff between local-global performance, while retaining desirable properties such as pole-safety and spherical-surface-distance preservation. Across five tasks spanning classification, regression, and image-augmented prediction, Slepian encodings outperform baselines and retain performance advantages across a wide range of neural network architectures.

Problem

• geographic representation
• localized modeling
• high-resolution
• spatial data
• machine learning

Innovation

• Slepian functions
• geographic representation
• localized encoding
• spherical harmonics
• pole-safety