UAV imagery and in-situ measurements for structure-from-motion snow depth mapping over the Laurichard rock glacier, France - surveyed in 2017

Published: 10 September 2019| Version 2 | DOI: 10.17632/9rhscd27y4.2


Unmanned aerial veichle (UAV) imagery and in-situ field measurements at the Combe de Laurichard, France (45.01ºN, 6.37ºE, 2500 m a.s.l.) were collected to explore uncertainties in mapping snow depth with structure-from-motion and multi-view stereo 3D reconstruction in an alpine area. Repeat UAV surveys were flown on each survey date to create multiple elevation models to determine the precision (i.e., repeatability) of SFM-MVS elevation models. This measure of uncertainty can be used to determine the precision of SFM-MVS snow depths using a model of error propagation. It also can illustrate how uncertainty in the SFM-MVS snow depths and elevation models vary spatially. The repeated snow-cover (snow-on) elevation models (6 in total) were acquired on June 1, 2017, and the snow-free (snow-off) elevation models (7 in total) on October 5, 2017. These elevation models were derived by performing SFM-MVS reconstruction using Agisoft PhotoScan. The UAV imagery was surveyed using a DJI Phantom 4, which flew in pre-programmed parallel flight paths with 75% side and top image overlap. The flying height of the UAV was approximately 60 m above ground level. Artificial targets were used for SFM camera calibration and georeferencing using the RGF93 / Lambert-93 projection and the NGF-IGN69 vertical datum (EPSG::5698). Validation data was collected by measuring topographic heights (i.e., check points) using a real-time-kinematic (RTK) global navigation satellite system (GNSS) survey with an accuracy < 2 cm (at 1 σ). At each check point location, snow depths were measured using an avalanche probe to a maximum 3 m depth. This RTK-GNSS survey was also used for collecting ground control points (GCPs). The position of the base-station was corrected using the PUYA reference station, which is located approximately 19 km from the study area. Additionally, to compare the uncertainty of SFM-MVS snow depths in stable and active deforming terrain (i.e., rock glacier creep), a mask of the rock glacier area was mapped using the UAV derived imagery and elevation models.



Structure from Motion, Digital Elevation Model, Remote Sensing of Ice and Snow, High Spatial Resolution, Snow, Drone (Aircraft)