Mapping Surface Temperatures on a Debris-Covered Glacier With an Unmanned Aerial Vehicle

Abstract

A layer of debris cover often accumulates across the surface of glaciers in active mountain ranges with exceptionally steep terrain, such as the Andes, Himalaya, and New Zealand Alps. Such a supraglacial debris layer has a major influence on a glacier's surface energy budget, enhancing radiation absorption, and melt when the layer is thin, but insulating the ice when thicker than a few cm. Information on spatially distributed debris surface temperature has the potential to provide insight into the properties of the debris, its effects on the ice below and its influence on the near-surface boundary layer. Here, we deploy an unmanned aerial vehicle (UAV) equipped with a thermal infrared sensor on three separate missions over one day to map changing surface temperatures across the debris-covered Lirung Glacier in the Central Himalaya. We present a methodology to georeference and process the acquired thermal imagery, and correct for emissivity and sensor bias. Derived UAV surface temperatures are compared with distributed simultaneous in situ temperature measurements as well as with Landsat 8 thermal satellite imagery. Results show that the UAV-derived surface temperatures vary greatly both spatially and temporally, with −1.4 ± 1.8, 11.0 ± 5.2, and 15.3 ± 4.7 • C for the three flights (mean ± sd), respectively. The range in surface temperatures over the glacier during the morning is very large with almost 50 • C. Ground-based measurements are generally in agreement with the UAV imagery, but considerable deviations are present that are likely due to differences in measurement technique and approach, and validation is difficult as a result. The difference in spatial and temporal variability captured by the UAV as compared with much coarser satellite imagery is striking and it shows that satellite derived temperature maps should be interpreted with care. We conclude that UAVs provide a suitable means to acquire surface temperature maps of debris-covered glacier surfaces at high spatial and temporal resolution, but that there are caveats with regard to absolute temperature measurement.

Assessing the potential of drone‐based thermal infrared imagery for quantifying river temperature heterogeneity

“…Indeed, a recent study by Abolt et al () reported similar issues of temperature drift with both the FLIR Tau 2 sensor (upon which the DJI Zenmuse XT camera used in this study is based) and also the FLIR Vue Pro sensor, indicating that this problem is both (a) relatively common and (b) not limited to one particular camera model. Similarly, recent work with sUAS‐TIR imaging of glacier temperatures revealed temperature drift of a comparable magnitude using a SenseFly ThermoMap TIR camera, albeit under very different environmental conditions (Kraaijenbrink et al, ). During the current study, we tested three versions of the same camera, which all generated very similar results, emphasising that observed drift was not an artefact of a single faulty camera.…”

“…It is possible that greater drift results from insufficient insulation (or shielding) from external influences in comparison with larger TIR cameras. However, we also hypothesise that camera miniaturisation necessitates closer mounting of electronic components, potentially resulting in increased heat build-up (e.g., Kraaijenbrink et al, 2018;Ribeiro-Gomes et al, 2017) and thus greater temperature drift in comparison with larger TIR cameras. This presents a continued challenge for sUAS-based TIR imagery acquisition.…”

“…This is potentially due to the use of a (relatively) wide angle lens, which is required when conducting flights at low altitude, increasing the angle of incidence at locations towards the edge of the lens's field of view (e.g., Kim, Park, Kopilevich, Tuell, & Philpot, 2013). Increased within-image biases may also result from vignetting, which is also associated with wider angle lenses (e.g., Goldman, 2010;Kelcey & Lucieer, 2012 (Kraaijenbrink et al, 2018). During the current study, we tested three versions of the same camera, which all generated very similar results, emphasising that observed drift was not an artefact of a single faulty camera.…”

Detecting Short-Term Surface Melt on an Arctic Glacier Using UAV Surveys

“…They are a particularly attractive tool for glaciological research due to their ability to collect data over wider areas and inaccessible regions of glaciers, with relatively low costs and efforts. As a result, there have been several recent glaciological studies relying on UAV data [7,[10][11][12][13][14][15][16][17][18]. Most commonly, these studies are using imagery in the visual spectrum from UAV-mounted digital cameras and SfM to recreate glacier surfaces and track motion.…”

“…Most commonly, these studies are using imagery in the visual spectrum from UAV-mounted digital cameras and SfM to recreate glacier surfaces and track motion. Glaciological studies using UAVs have tracked glacier motion (e.g., [7,10,14]); measured seasonal or multi-year melt (e.g., [7,10,13]); analyzed calving and crevasse dynamics (e.g., [12,15,16]); and mapped surface characteristics such as drainage networks [11], albedo [17], debris cover [18], and cryoconite holes [19].…”