Item: Consistency and Accuracy of Public Avalanche Forecasts in Western Canada
-
-
Title: Consistency and Accuracy of Public Avalanche Forecasts in Western Canada
Proceedings: International Snow Science Workshop Proceedings 2018, Innsbruck, Austria
Authors:
- Grant Statham [ Parks Canada Agency, Banff, AB, Canada ] [ Simon Fraser University, Burnaby, BC, Canada ]
- Stephen Holeczi [ Parks Canada Agency, Banff, AB, Canada ]
- Bret Shandro [ Simon Fraser University, Burnaby, BC, Canada ]
Date: 2018-10-07
Abstract: Forecasting avalanche problems and avalanche danger is a judgmental assessment that is highly susceptible to interpretation and bias. Thus, congruency between forecasters is a significant challenge and much effort has been expended to harmonize assessment methods between different forecasters, regions, and even countries. While recent studies have helped to identify bias and inconsistency in avalanche danger ratings (Lazar et al. 2016; Techel et al. 2018), in-house feedback directly to forecasters is sometimes absent. Accurate, well-summarized feedback could provide the primary basis for avalanche forecasters to improve their judgmental forecasts. Using data from historic forecasts, this paper looks first at the consistency of how avalanche problems are applied by different agencies with adjacent regions in the Canadian Rockies. We show inconsistency in the distribution of avalanche problems published for adjacent regions with similar snowpacks, and by different forecasters within the same region. Although definitions exist for different types of avalanche problems (Statham et al. 2018), insufficient guidance for forecasters on how to apply avalanche problems consistently can lead to conflicting information and confusion for backcountry users. Next we look at the accuracy of 24, 48 and 72 hour forecasts of danger ratings when compared against real-time assessments. Drawing from 3,752 avalanche bulletins over seven seasons, we show an overall accuracy of 73%. Forecasts of Low danger are the most accurate (84%) and they become progressively less accurate as the forecast danger levels rise. We conclude by offering recommendations on the application of avalanche problems, enhanced forecaster training and encouragement for other agencies to analyze their own forecasting data. Feedback can have a pronounced effect on bias if incorporated more routinely into professional activities (Vick 2002) and with it, forecasters can become better calibrated.
Object ID: ISSW2018_O17.1.pdf
Language of Article: English
Presenter(s):
Keywords: avalanche problems, danger, forecasting, consistency, accuracy
Page Number(s): 1491-1495
-