Скачать или смотреть Map the Gap: Detecting and Characterizing Pan-Arctic Infrastructure with Satellite Imagery and AI

Elias Manos & Chandi Witharana, University of Connecticut. PDG Webinar May 2025.

Permafrost thaw threatens to disrupt the lives of millions of people living in the Arctic by damaging homes, roads, and various other critical buildings and structures needed to sustain over 1000 Arctic communities on permafrost. Impacts are already unfolding as observed across the region. However, there is a lack of unified, high-resolution infrastructure data that is comprehensive in geographic coverage and attribute information across the pan-Arctic region. This data is critical to accurately estimate economic losses due to permafrost thaw damages, which is needed on the international scale to track disaster risk reduction goals, and on local to regional scales to prepare hazard mitigation plans and acquire the necessary resources. Using artificial intelligence models trained on geospatial data, we can bridge this gap by: (1) detecting infrastructure such as buildings and roads from Maxar satellite imagery, (2) predicting attributes of buildings, such as use type (i.e., residential vs. non-residential building). Findings show how this work has filled geographic gaps in the best available pan-Arctic building dataset (OpenStreetMap) and has the potential to fill in its largely missing building attribute information.