Using Low Cost Commercial UAVs to Measure Snow Depth and Surface Change

Small commercial unmanned aerial vehicles (UAVs) provide a unique opportunity for the remote sensing of snow at high spatial resolutions and relatively low cost. Aerial imagery acquired via UAV can be used to develop three-dimensional surface models by using structure from motion (SfM) methodologies. Surface models developed over monthly, weekly, daily and hourly time frames can be compared to analyze differences in snow depth and study snow dynamics related to melt and wind processes. Additionally, the mobility provided by UAVs can reduce the impact of traditional obstacles to snow remote sensing such as complex topography and vegetation. Multiple flights were conducted at two field sites on Grand Mesa, CO using a DJI Phantom 4 UAV. These selected sites included different land cover types including open grass, rocky terrain, and alpine forest. Multiple missions were flown on sequential days to monitor snow depth changes due to snow melt and other processes. Image acquisition occurred at 50 meters above ground level to produce digital elevation models with a spatial resolution of two centimeters. A differential GPS system and ground control point targets were employed to attain an absolute geolocation accuracy of approximately two centimeters. Traditional snow depth surveys were conducted at both sites to establish a ground truth baseline for accuracy assessment of the snow surface models. Preliminary comparisons between the UAV produced snow depth models and traditional snow survey depths show a median difference of 8.1 centimeters and an RMSE of 17 centimeters.