Landform mapping, elevation modelling, and thaw subsidence estimation for permafrost terrain using a consumer-grade remotely-piloted aircraft

• Published in: Drone Systems and Applications
• Main Authors: Oldenborger, Greg A., Bellehumeur-Génier, O., LeBlanc, Anne-Marie, McMartin, I.
• Format: Article in Journal/Newspaper
• Language: English
• Published: Canadian Science Publishing 2022
• Subjects: Nunavut; Rankin Inlet; permafrost
• Online Access: http://dx.doi.org/10.1139/dsa-2021-0045; https://cdnsciencepub.com/doi/full-xml/10.1139/dsa-2021-0045; https://cdnsciencepub.com/doi/pdf/10.1139/dsa-2021-0045

We assess performance of a small consumer-grade remotely-piloted aircraft (RPA) for landform mapping, elevation modelling, and thaw subsidence estimation in continuous permafrost terrain. We acquired RPA imagery near Rankin Inlet, Nunavut, to construct orthomosaics and digital elevation models (DEMs) that we use to interpret geomorphology and surficial geology. We estimate seasonal thaw subsidence using DEM differences. To quantify accuracy, RPA DEMs are compared with a satellite-based reference elevation. Subsidence estimates are compared with measurements from differential interferometric synthetic aperture radar (DInSAR). We find that RPA images are very effective for mapping periglacial landforms and surficial geology with the chosen flight specifications. The DEMs exhibit vertical mean absolute error of approximately 1 cm at ground control points. Away from control points, relative vertical accuracy is approximately 3 cm. Comparison to the reference elevation results in survey-wide vertical mean absolute errors of 33–66 cm with high variability and spatial autocorrelation of elevation discrepancy. There is local agreement between DEM differences, DInSAR, and on-the-ground measurements of seasonal subsidence. Results suggest that small RPA may be applicable for mapping thaw subsidence on the order of a few centimetres near control points. However, DEM differences are influenced by vegetation and are contaminated by spatially-variable artefacts, preventing reliable survey-wide RPA estimation of seasonal thaw subsidence.