Quantifying Snow Volume Uncertainty from Repeat Terrestrial Laser Scanning Observations

Abstract:

Terrestrial laser scanning (TLS) systems are capable of providing rapid, high density, 3D topographic measurements of snow surfaces from increasing standoff distances. By differencing snow surface with snow free measurements within a common scene, snow depths and volumes can be estimated. These data can support operational water management decision-making when combined with measured or modeled snow densities to estimate basin water content, evaluate in-situ data, or drive operational hydrologic models. In addition, change maps from differential TLS scans can also be used to support avalanche control operations to quantify loading patterns for both pre-control planning and post-control assessment. However, while methods for computing volume from TLS point cloud data are well documented, a rigorous quantification of the volumetric uncertainty has yet to be presented. Using repeat TLS data collected at the Arapahoe Basin Ski Area in Summit County, Colorado, we demonstrate the propagation of TLS point measurement and cloud registration uncertainties into 3D covariance matrices at the point level. The point covariances are then propagated through a volume computation to arrive at a single volume uncertainty value. Results from two volume computation methods are compared and the influence of data voids produced by occlusions examined.