Publication Date

12-2014

Date of Final Oral Examination (Defense)

10-17-2014

Type of Culminating Activity

Dissertation

Degree Title

Doctor of Philosophy in Geophysics

Department

Geosciences

Supervisory Committee Chair

Hans-Peter Marshall, Ph.D.

Supervisory Committee Member

John Bradford, Ph.D.

Supervisory Committee Member

Alejandro N. Flores, Ph.D.

Supervisory Committee Member

Jeffrey B. Johnson, Ph.D.

Abstract

Avalanche formation is a complex interaction between the snowpack, weather, and terrain. However, detailed observations typically can only be made at a single point and must be extrapolated over the slope or regional scale. This study aims to provide avalanche forecasters with tools to evaluate the snowpack, avalanche hazard, and avalanche occurrence when manual observations are not feasible.

Avalanches that occur within the new storm snow are a prevalent problem for the avalanche forecasters with the Idaho Transportation Department (ITD) along Highway 21. We have implemented a real time SNOw Slope Stability (SNOSS) model that provides an index to the stability of that layer. SNOSS has been run real time starting during the winter of 2011/2012 with model results outputted to a webpage for easy viewing by avalanche forecasters.

To further improve the accuracy of SNOSS, the model was evaluated with a large database of avalanches from the Utah Department of Transportation (UDOT). Using weather data and SNOSS results, the probability of an avalanche day producing a natural direct action avalanche was calculated using a Balanced Random Forest (BRF). In the future, we hope that the BRF can provide a probability of an avalanche occurrence given the current weather and snowpack conditions that can be utilized by avalanche forecasters in their normal operations.

The concern for avalanche forecasters with highway operations is the threat of an avalanche releasing and hitting a highway. Infrasound generated by an avalanche moving downhill can be detected and tracked using array processing techniques. This will allow avalanche forecasters to evaluate the avalanche hazard more effectively by determining when and where avalanches have occurred. An avalanche detection system has been developed to detect avalanches in near real time using infrasound arrays. The system processes the infrasound data on-site, automatically detects events, and classifies the events using multiple neural networks. If an avalanche has been detected, the system will transmit the necessary information over satellite to be viewed by avalanche forecasters on a webpage.

Share

COinS