- Abstract
- The natural and built environments of the Arctic permafrost regions are experiencing rapid transformations due to a warming climate, necessitating the development of novel monitoring tools to accurately map permafrost landforms and thaw disturbances, track their changes over time, and to assess the economic impact on human-built infrastructure. The entire Arctic has been imaged by very high spatial resolution (VHSR) Maxar satellite sensors at the sub-meter resolution multiple times during the last two decades, capturing dynamics of individual microtopographic features, thaw disturbances, and human-built infrastructure features without sacrificing geographic extent and spatial detail. While data repositories holding millions of these VHSR images have reached the petabyte scale, imagery-derived pan-Arctic scale science products are yet rare, naturally setting the stage for artificial intelligence (AI) algorithms like the deep learning (DL) convolutional neural network (CNN). However, scalability of automated analysis over millions of square kilometers comprising heterogeneous Arctic landscapes is a challenging task, requiring the development of efficient image-to-assessment workflows that center on high performance computing resources. Therefore, we present here such a novel image analysis tool, High-resolution Arctic Built Infrastructure and Terrain Analysis Tool (HABITAT), that enables the integration of operational-scale GeoAI capabilities into Arctic science applications, with four example mapping scenarios (ice-wedge polygons, ice-wedge polygon troughs, retrogressive thaw slumps. and human-built infrastructure).
- Presented by
- Elias Manos
- Institution
- (1) Department of Natural Resources and the Environment, University of Connecticut; (2) Woodwell Climate Research Center, Falmouth, MA, USA
- Keywords
- Remote sensing, Artificial intelligence, Permafrost thaw, Ice-wedge polygons, Infrastructure