UAV-borne GPR for snowpack characterization: Potential, limitations and operational guidelines

Abstract

Characterization and monitoring of the snow cover is highly applicable for numerous problems in mountain and arctic environments, including avalanche hazard and hydrology assessments. Remote sensing techniques from space or air are conventionally used to map and monitor large areas. To bridge the gap between satellite and ground scales, we propose the use of UAV-borne GPR systems to optimize areal coverage, resolution and repeatability for snowpack information. We present an extensive study of the potential applications and limitations when using UAV-borne GPR for snowpack characterization. We discuss the operational constraints and demonstrate various examples conducted in different environments. We tested low (400 MHz) and high (1 GHz) frequency GPR systems on a commercial off-the-shelf UAV equipped with a radar altimeter and terrain following capabilities. GPR data were acquired at three field sites in central and western Norway. We demonstrate that data repeatability is satisfactory and that any measurement differences are due to aircraft positioning errors. We also tested data acquisition from various flight altitudes above the snow surface which led us to the conclusion that recording data between altitudes of 2 to 4 m above the surface is the best compromise between flight safety and data quality. Regarding flight speeds, if we focus on layer tracking and do not require high lateral resolution for discrete target mapping, data can be acquired at 2 m/s or more as long as flight safety is ensured. We recommend recording data along pre-determined flight paths, downslope starting from the highest elevation and following profiles parallel to the slope. Kinematic timelapse surveys recorded at a two-week interval highlighted the capabilities for thin layer detection. Centimeter thick melt-freeze crust layers are visible on GPR profiles and are correlated with observed layers in neighboring snowpits. Additional tests in wet snowpacks showed that the penetration depth was limited with the 1 GHz antenna due to the attenuation related to the presence of liquid water in the snowpack. The use of a lower frequency antenna (400 MHz) enabled sufficient penetration depth even though it caused a loss in resolution for mapping the upper layers of the snowpack. Overall, the UAV GPR surveys showed promising results towards recording highly repeatable snow height and snow layering data which can be helpful for avalanche forecasting and hydrology studies.