Potential of Balloon Photogrammetry for Spatially Continuous Snow Depth Measurements

Abstract

We carried out two aerial surveys using a weather balloon as the platform to measure the snow depth in the Wolverton watershed, CA, USA: one when the site was snow covered and the other one after the snow melted out. We reconstructed the 3-D surfaces of the site using the structure-from-motion (SfM) photogrammetry of the photographs taken in the surveys and differenced the heights of the two surfaces to obtain the snow depth. The snow depth estimates corresponded well with 32 manual measurements of snow depth, with R = 0.87 (p <; 0.05) and a root-mean-square error (RMSE) of 7.6 cm, the majority of which is a 6-cm systematic bias due to the vegetation rebound in the snow-off measurements. The relative depth error is 17% in the extremely dry year of sampling (i.e., 2015) and is expected to decrease for deeper snow because the absolute error of SfM is relatively static. The processed snow depth is able to capture the snow spatial variability at submeter scale. This study suggests that balloon photogrammetry is a repeatable, flexible, economical, and safe method for continuous snow depth measurement at small scales and could complement existing remote sensing platforms (e.g., aircrafts, satellites, and drones) for snow observations in open areas by providing spatial continuity, long observation time, and customizable resolution.