Development of pipeline to measure snowpack evolution through the melting period using SLAM-Enabled Mobile Terrestrial LiDAR

Abstract:

In mountainous regions of western North America the snowpack is the main driver of stream flows. Removal of forests either through logging or by natural disturbance such as forest fire has the potential to alter processes of snow accumulation and melt and, as well, alter the stream flows of these mountain watersheds. In the past, the analysis of the effect of forest clearing on snow accumulation and melt processes required labour intensive, manual surveys of snow depth and density. LiDAR (Light Detection and Ranging) is rapidly evolving as a tool for natural resource management. Handheld SLAM LiDAR devices have been shown to be very effective for measuring snow depth over time under a forest canopy. The device also allows for the acquisition of forest metrics, such as tree height, stem density and basal area within forest stands. However, analysis of large LiDAR datasets remains an issue. Georeferencing massive point clouds and extracting the surface models one scan at a time is a cumbersome task. The objective of the Rover Pipeline project was to develop a more effective way to process LiDAR data for the purpose of quantifying forestry effects on snow accumulation and melt across multiple stands in a mountainous watershed.

During the 2019 spring snowmelt period a study was initiated in Rover Creek, a watershed situated in the southern Selkirk Mountains of British Columbia, to investigate the use of LiDAR to quantify the effects of logging on snow accumulation and melt processes. During the 2020 snowmelt period the study covered thirteen stands that were situated across a range of elevations and aspects in the watershed. LiDAR data was collected throughout the melt period at intervals ranging from weekly to every few days using a handheld GeoSLAM Zeb Horizon LiDAR unit. A programming ‘pipeline’ was developed to process the data and extract the snow surface model from the point clouds. A comparison of ground classification algorithms available within open source software was also accomplished. Development of the processing pipeline has dramatically improved efficiency in analysing the point cloud data whilst generating accurate and very high-resolution snow depth grids of the snowpack beneath the forest stands. Future work using manned LiDAR is planned for the upcoming winter season to improve data acquisition efficiency.