Browsing by Author "Kucharczyk, Maja"

Now showing 1 - 3 of 3
  • Thumbnail Image
    ItemOpen Access
    Pre- and Post-Disaster Remote Sensing with Drones for Supporting Disaster Management
    (2023-04-27) Kucharczyk, Maja; Hugenholtz, Chris H.; Geldsetzer, Torsten; Hay, Geoffrey J.; Moorman, Lynn; Slick, Jean
    Small (< 25 kg) aerial drones have expanded the remote sensing toolkit for disaster management activities, resulting in hundreds of published case studies in the past two decades. The overall goal of this doctoral research, which comprises three related studies, is to evaluate drone-based pre- and post-disaster remote sensing as a tool to support disaster management. The first study provides a critical review of drone-based remote sensing in natural hazard-related disasters to highlight research trends, biases, and expose new opportunities. Recommendations for future research include a greater focus on demonstrating and evaluating drone-based support of pre-disaster data acquisition (a preparedness activity) and rapid damage assessment (a response activity). As such, the second study presents the first pre-disaster drone-based mapping mission over an urban area (downtown Victoria, British Columbia) approved by Transport Canada. The objective was to assess the quality of 3D data obtained with the only legally approved drone. Finally, the third study demonstrates rapid mapping of hurricane roof damage using artificial intelligence (deep learning) and drone imagery, including an accuracy assessment. Overall, this doctoral research identified critical knowledge gaps in the field of pre- and post-disaster remote sensing with small aerial drones, and then demonstrated and evaluated drone-based support of two building-damage-related activities that were recommended for future research.
  • Thumbnail Image
    ItemOpen Access
    Reported UAV incidents in Canada: analysis and potential solutions
    (NRC Research Press, 2017-06-01) Nesbit, Paul R.; Barchyn, Thomas E.; Hugenholtz, Chris H.; Cripps, Sterling; Kucharczyk, Maja
    UAV incidents were analyzed using data from Transport Canada's Civil Aviation Daily Occurrence Reporting System (CADORS). Between 05 November 2005 and 31 December 2016 a total of 355 incidents were reported in Canadian airspace. The largest number involved UAV sightings (66.5%) and close encounters with piloted aircraft (22.3%). These incidents increased markedly after 2013, with the highest number in British Columbia, followed by Ontario, Quebec, Alberta, and Manitoba. The vast majority of UAV incident reports were filed by pilots of piloted aircraft. Typically, airspace at altitudes greater than 400 feet above ground level (AGL) is off limits to UAVs; however, of the 270 incidents in the CADORS database with UAV altitude reported, 80.4% were above 400 feet AGL and 62.6% were above 1000 feet AGL. Of the 268 incidents with reported horizontal distance to the nearest aerodrome, 74.6% occurred or likely occurred within 5 nautical miles (nm), and of those 92.4% and 76.6% were reported above 100 and 300 feet AGL, respectively. Collectively, the CADORS data indicate that the overwhelming majority of UAV incidents reported in Canada were airspace violations. These results can guide future risk mitigation measures, hardware/software solutions, and educational campaigns to increase airspace safety.
  • Thumbnail Image
    ItemOpen Access
    UAV-LiDAR and Structure from Motion Photogrammetry: Spatial Accuracy in Vegetated Terrain
    (2017) Kucharczyk, Maja; Hugenholtz, Chris; Hall-Beyer, Mryka; Lichti, Derek
    Multiview stereo images acquired by uninhabited / unmanned aerial vehicles (UAVs) in combination with structure from motion (SfM) photogrammetry have created new capacity to develop high-resolution geospatial data, but vertical error is typically higher in vegetated areas because the ground surface is not visible in stereo. Miniaturized LiDAR systems for UAVs have potential to overcome this limitation, but their vertical accuracy in different vegetation types is not well documented. This thesis evaluated the accuracy of UAV-LiDAR and UAV-SfM in six vegetation types: grasses (short and tall), shrubs (short and tall), and trees (deciduous and coniferous). Results indicate UAV-LiDAR was more accurate in estimating ground elevation in all types, while vegetation height accuracy was higher for some types with UAV-SfM. UAV-LiDAR consistently sampled sub-canopy tree structure, while UAV-SfM only captured tree tops. Several factors are proposed to explain these differences and direct future research.