Remote sensing of subglacial bedforms from the British Ice Sheet using an Unmanned Aerial System (UAS): Problems and Potential.

• Main Author: CLAYTON, ALEXANDER,IAN
• Format: Thesis
• Language: unknown
• Published: 2012
• Subjects: Glacial; Bedforms; Remote sensing; UAV; UAS; British Ice Sheet; Drumlin; Pams; Ice Sheet
• Online Access: http://etheses.dur.ac.uk/5907/; http://etheses.dur.ac.uk/5907/1/MScR_Thesis_AClayton.pdf

Photogrammetry can be applied to the results of UAS (Unmanned Aerial Systems) based photographic surveys to produce high resolution DEMs (Digital Elevation Models) of small areas (c. 1 km2). However, this method has not been widely used in academia due to photogrammetric programmes working poorly with the ill constrained intrinsic and extrinsic properties that often accompany UAS based photographs. In this study a PAMS (Personal Aerial Mapping System) SmartOne B UAS was used to provide image sets for testing a number of different photogrammetry packages; LPS, Bundler, PhotoSynth and PhotoScan, with the aim of producing sub-metric accuracy DEMs with a low complexity methodology and without significant financial investment. To demonstrate the potential use of a UAS photogrammetric survey methodology it was applied here to an investigation into scale dependant remote sensing of glacial geomorphology. Subglacial bedforms, landforms produced by the flow of ice over land, are thought to ‘seed’ with a minimum horizontal dimension of 100 m. This hypothesis is based on surveys of bedforms across the UK and Ireland using NEXTMap DEMs with 1 m accuracy and 5 m resolution. Here we test that hypothesis using sub-metric accuracy DEMs produced via photogrammetry of an area in the Eden Valley drumlin field, NW England. The UAS was found to be suitable for this type of survey, but only one of the four photogrammetry programmes provided an effective and low complexity methodology. This programme, PhotoScan, was shown to require minimal user training and could produce DEMs from the survey imagery on the day of flying with a standard high performance computer at a resolution of 0.12 m2. The DEM produced was down sampled and validated against pre-existing 1 m LiDAR (Light Detection And Ranging) data of the same area. It showed poor absolute accuracy due to a systematic parabolic error introduced during processing that made quantification of the DEM error problematic. However, estimates of the error additional to this systematic ...