Harnessing multi-source geospatial big data to construct a pan-Arctic building footprint atlas: integrating geometric, morphological, temporal, and functional attributes
Year: 2026
Type: article
Abstract: Arctic infrastructure is increasingly exposed to permafrost degradation and thaw-related geohazards, yet building-resolved information remains insufficient for translating regional climate risk into object-level exposure assessments. Existing settlement, impervious-surface, and land-cover products delineate broad built-up extent but lack the object-level detail needed to characterize individual Arctic buildings. Here, we construct a harmonized multi-attribute pan-Arctic building footprint atlas by integrating complementary building-footprint products with information from multiple geospatial sources. The three input products were evaluated against 39,685 manually interpreted reference buildings across 12 Arctic benchmark sites, then integrated through hierarchical geometric conflation, redundancy removal, and multi-source attribute enrichment. Independent validation against regional reference building footprints yielded an overall area-based F1-score of 77.3% for the final fused atlas. The resulting atlas identifies approximately 15.08 million buildings north of 60°N, covering about 2,004 km2, and provides object-level geometry together with available morphological, temporal, and functional attributes. The atlas reveals a highly uneven Arctic built environment, with dense concentrations along the southern Arctic margin, particularly in northern Europe, western Russia, southern Alaska, and Iceland, whereas the high Arctic remains sparsely built. Linking the atlas with environmental layers shows that, among footprints with valid permafrost assignments, 31.7% of footprint area and 21.9% of buildings occur in locations with permafrost probabilities of ≥50%. Under Representative Concentration Pathway 8.5, high-hazard-potential zones contain 51% of the footprint area with valid hazard assignments by 2061–2080. The Pan-Arctic Building Footprint Atlas v1 is publicly available via https://doi.org/10.5281/zenodo.20796872.
Source: Figshare
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7234809590
