Arctic circumpolar permafrost region building footprints from <1 meter resolution Maxar satellite imagery and OpenStreetMap Version 2 (2018-2023)

This product is a geospatial vector layer containing two-dimensional footprints of buildings (i.e., spatial extent covered by an individual building on the ground) across the Arctic circumpolar permafrost region. The data is based on building footprints from OpenStreetMap (OSM) contributions within Arctic regions, then built upon by filling in missing areas with building footprints detected from less than 1-meter spatial resolution, summertime, cloud-free Maxar satellite imagery of Arctic circumpolar permafrost communities. The dataset was produced using the HABITAT (High-resolution Arctic Built Infrastructure and Terrain Analysis Tool) deep learning framework, which consists of a convolutional neural network (CNN) segmentation model and a GraphCNN classification model, as well as post-processing and quality control (vectorization, smoothing, false positive + small footprint removal). Building heights and number of stories are estimated based on the ArcticDEM circumpolar digital surface model. Building footprints are provided in the North Pole Lambert Azimuthal Equal Area projection. The provided data spans 32 first-level administrative regions (the largest subnational administrative unit within a country) spanning the Arctic: Alaska (US), Yukon, Northwest Territories, Nunavut, Newfoundland and Labrador, Northern Quebec (Canada), all Greenland regions, all Iceland regions, Nordland, Troms, Finnmark, Svalbard (Norway), Norbotten (Sweden), Lappi (Finland), Komi, Arkhangelsk, Nenets, Khanty-Mansi, Yamalo-Nenets, Krasnoyarsk Krai, Sakha Republic, Kamchatka, Magadan, and Chukotka (Russia).