Data analytics and software to support avalanche forecasting decisions

Abstract

Avalanches are a very powerful force of nature and pose significant risk for ski areas and mountainous roads. Avalanche forecasting and mitigation are a very important part of keeping the public safe. Terrestrial laser scanning lidar systems have proven useful in more accurate forecasting and mitigation efforts, but utilizing them can be time consuming. The goal of this project is to operationalize a workflow and create algorithms and ultimately produce a software product that can rapidly analyze snow covered mountainous terrain, allowing avalanche forecasters to make informed decisions on where to focus their mitigation efforts. In this dissertation, I first present algorithms that were designed to align scans, identify trees and cliffs, grid scans, and calculate snow depth. I then introduce a software package that was implemented incorporating these algorithms with a point cloud visualization tool. This software package allows a user to control and visualize the analysis process to make more informed avalanche mitigation decisions. Algorithms were parameterized and validated with a field study consisting of data collection events at Bridger Bowl, Bear Canyon, and the Yellowstone Club in Montana. A Riegl VZ-6000 TLS lidar system was used for all data collection efforts. This dissertation documents the design of this analytics workflow by presenting the algorithms developed, discussing the software implemented, and presenting the data collection efforts that guided the design of the algorithms and served to validate their efficacy.