Item: METEOROLOGICAL AND SNOWPACK PROPERTIES ASSOCIATED WITH CRUST-ADJACENT PERSISTENT WEAK LAYERS. PART 2: APPLYING THEORY TO OBSERVED PATTERNS
-
-
Title: METEOROLOGICAL AND SNOWPACK PROPERTIES ASSOCIATED WITH CRUST-ADJACENT PERSISTENT WEAK LAYERS. PART 2: APPLYING THEORY TO OBSERVED PATTERNS
Proceedings: International Snow Science Workshop 2024, Tromsø, Norway
Authors:
- Andrew Schauer [ Chugach National Forest Avalanche Center ]
- Andy Moderow [ Chugach National Forest Avalanche Center ]
- Aleph Johnston-Bloom [ David Hamre and Associates ]
Date: 2024-09-23
Abstract: Over the past nine seasons, ten substantial melt-freeze crusts formed in avalanche start zones along Turnagain Pass, in the heart of Alaska’s Chugach National Forest. Near-crust faceting was frequently observed after crusts were buried, producing a persistent weak layer that would often become a forecasting challenge for weeks or months to come. Massive dry and wet slab avalanche cycles and dangerous near-misses were credited to the facets that developed above, within laminations of, or below these crusts. However, not all crusts developed a weak layer that would produce long-term avalanche activity. This study explores weather and snowpack observations that may help avalanche practitioners anticipate long-term avalanche problems on crust-adjacent persistent weak layers. A companion paper reviews existing research related to crusts and persistent instability, identifying long- and short-term cumulative loading, temperature trends, and stability test results as potential indicators of problematic crust-adjacent weak layers. Here, we explore the practical utility of applying these factors as predictors of crust-related avalanche activity by comparing them against two groupings of well-documented crusts at Turnagain Pass: (1) those with a weak layer that ultimately produced persistent avalanche activity, and (2) those with short-lived or no reactivity. We use Support Vector Machines to identify loading and temperature thresholds that separate active and inactive seasons. We then use a simple contingency table analysis to explore the best ways to apply stability test results and individual loading event magnitude toward predicting avalanche activity. This work highlights the utility of previously-identified predictors, with an emphasis on putting theory into practice. We also identify areas for potential future work along these lines, and ways to improve current practices to improve our ability to anticipate avalanche activity on crust-adjacent persistent weak layers.
Object ID: ISSW2024_P1.14.pdf
Language of Article: English
Presenter(s): Andrew Schauer
Keywords: crust, persistent weak layer, machine learning
Page Number(s): 160 - 167
-