Item: UAV-BORNE GROUND PENETRATING RADAR FOR AVALANCHE FORECASTING
-
-
Title: UAV-BORNE GROUND PENETRATING RADAR FOR AVALANCHE FORECASTING
Proceedings: International Snow Science Workshop 2024, Tromsø, Norway
Authors:
- Bastien Dupuy [ SINTEF, Trondheim, Norway ]
- Martin Chatel [ SINTEF, Trondheim, Norway ] [ Ecole Polytechnique, Paris, France ]
- Arnt Grøver [ SINTEF, Trondheim, Norway ]
- Stéphane Garambois [ Université Grenoble-Alpes, CNRS, IRD, Univ. Savoie Mont Blanc, ISTerre, Grenoble, France ]
- Madeline Lee [ SINTEF, Trondheim, Norway ]
- Andrew Tobiesen [ SINTEF, Trondheim, Norway ]
- Halgeir Dahle [ Norwegian Public Roads Administration, Molde, Norway ]
- Sean Salazar [ Norwegian Geotechnical Institute, Oslo, Norway ]
- Regula Frauenfelder [ Norwegian Geotechnical Institute, Oslo, Norway ]
- Jordy Hendrikx [ The Arctic University of Norway UiT, Institute of Geosciences, Tromsø, Norway ] [ Antarctica New Zealand, Christchurch, New Zealand ]
- Tore Humstad [ Norwegian Public Roads Administration, Molde, Norway ]
Date: 2024-09-23
Abstract: In the GEOSFAIR (Geohazard Survey from Air) innovation project for the Norwegian public sector, several UAV (Uncrewed Aerial Vehicles)-borne remote sensing payloads (LiDAR scanners; RGB, infrared, and multispectral cameras) were tested to gather information on the snow surface and its changes over time. For example, by comparing bare-earth and snow surface digital elevation models over time, it is possible to derive estimates of snow height and snow height changes. However, in addition to resolution limitations for snow height (which is dependent on the quality of baseline terrain models), none of these remote sensing methods can give information on the snow layering and snowpack properties. During the last three years, we tested UAV-borne GPR (Ground Penetrating Radar) sensors to get information on the subsurface of the snow, i.e., on snow height, snow layering and snow properties (snow density and liquid water content). Numerous field tests in real mountain conditions including mapping flat areas and BVLOS flights in avalanche starting zones enabled us to determine optimal operational flight guidelines. To optimize data quality, the GPR sensors should record data at an altitude less than 5 m above the snow surface, which requires altimeter and UAV terrain-following capabilities. Flying downslope at a speed up to 2-3 m/s and following the surface at 2-4 m have been shown to be the best compromise in terms of flight safety and data quality. In dry snowpacks, we use a shielded antenna with a 1 GHz central frequency, that penetrates up to 8-10 m of snow and can detect changes in layering down to a couple of centimeters. We show good correlations between snow pits and interpreted layers in the GPR data, especially when mapping melt-freeze crusts. By converting snow density to GPR wave velocity at a snow pit location, and by mapping snow surface and snow-ground interfaces, we are also able to derive high-resolution snow height maps which correlate well with snow height derived from LiDAR surveys with lower spatial resolution. UAV-borne GPR is a promising tool to provide remote snowpack information including high resolution snow height and snow layering and may be used to support local avalanche forecasting. Additional work is on-going to derive snow properties without the need for local snow pit information.
Object ID: ISSW2024_O8.7.pdf
Language of Article: English
Presenter(s): Bastien Dupuy
Keywords: remote sensing, GPR, UAV, snow height, snow density
Page Number(s): 1035 - 1040
-