Is It a Problem?

Exploring Avalanche Problem Assessments in Public Avalanche Forecasting

By Heather Hordowick and Pascal Haegeli

This article was originally published in The Avalanche Journal, Vol. 132, spring 2023

SINCE THE ADOPTION of the Conceptual Model of Avalanche Hazard (CMAH; Statham et al., 2018), avalanche problems have formed a fundamental component of avalanche hazard assessment and communication in Canada and beyond. However, the pathway from observations to avalanche problems is not explicitly defined in the model; therefore, these assessments rely heavily on subjective judgements that are prone to noise and bias. To shed light on these practices, the Simon Fraser University Avalanche Research Program (SARP) has undertaken several studies to better understand the application of the CMAH over the last few years.

In the first project, Moses Towell (2019) used a statistical approach to explore the relationship between the problems posted in the avalanche bulletin in Glacier National Park and simulated weather and snowpack observations at a representative location in the region. The results of this study confirmed trends that we would expect to see, such as a strong relationship between amounts of new snow and storm slab problems being added to the bulletin. However, the results were not so clear for other forecaster decisions. For example, no statistically relevant explanations were found linking simulated snowpack and weather data to the removal of persistent avalanche problems. This highlighted that there must be other factors at play that determine forecasters’ avalanche problem choices.

To dig deeper, we conducted a qualitative research project on avalanche problems in which we interviewed experienced forecasters about their personal practices for adding, transitioning, and removing avalanche problems from the public avalanche bulletin. Over the 2020-21 winter season, we conducted 22 1.5-hour interviews with forecasters from four different Canadian forecasting agencies: Avalanche Canada; Banff, Yoho, and Kootenay National Parks; Glacier National Park; and Kananaskis Provincial Park. In our interviews, we discussed in detail the assessment of either storm slab and wind slab problems, or persistent and deep persistent slab problems. To document our conversations, we used a method called concept mapping that has been applied extensively in the field of cognitive science to capture and describe expertise. The concept-mapping interviews resulted in visualizations that identified considerations and linked them together in a semi-hierarchal structure that represented each forecaster’s perspectives on a specific scenario such as ‘removing a storm slab problem’ (Fig. 1).

THE PHYSICAL PREDICTORS YOU MIGHT EXPECT

The CMAH distinguishes between nine avalanche problem types that are defined by their typical physical characteristics, formation, evolution and persistence, informative observation types, and effective mitigation options (Statham et al., 2018). As one might expect, these physical characteristics and observations were reflected in the concept maps from our interviews.

While the forecasters discussed a wide range of observation types for each scenario, generally all or nearly all the forecasters agreed on the relevance of a few key observation types, which was not surprising. For example, when discussing adding a storm slab problem, all forecasters referenced new snow, wind, and air temperatures as key predictors. However, eight additional observation types were mentioned by a majority of the forecasters, and another 18 observation types related to instability, snowpack, spatial, temporal, and weather factors were mentioned by at least two forecasters.

Comparing scenarios for adding a problem to those for removing the same problem showed that observation types for adding the problem tended to be more numerous and have a closer alignment with both the CMAH and the data-driven relationships from Towell’s (2019) study. This result reflects the challenge and lack of guidance associated with decisions about removing problems. For example, the predictors for adding a deep persistent slab problem mentioned by the forecasters were in line with the CMAH definitions related to the weak layer, slab hardness, persistence time, and avalanche size. The predictors for removing a deep persistent slab problem, on the other hand, only shared avalanche size as an observation type included in the CMAH definitions. However, since avalanche size is unlikely to decrease when considering removing the problem, the CMAH definitions do not provide practicable guidance for the removal of a deep persistent slab problem.

Focusing on the observation values related to each observation type revealed where forecasters agreed and where differences exist. In some cases, we observed divergences that can be related to differing snow climate or terrain between forecast regions. For example, when
asked about adding storm slab problems, forecasters from the Canadian Rockies (Banff and Kananaskis) had a lower limit for typical wind speeds than forecasters in transitional snow climates (Glacier National Park). This might be related to the fact that the less dense snow typical of a continental snow climate might be redistributed at lower wind speeds than in a transitional snow climate. However, physical differences between forecast regions did not provide logical explanations for all the differences observed.

THE ADDITIONAL CONSIDERATIONS YOU MIGHT NOT EXPECT

Going beyond the physical, snow-science-based aspects discussed above and formally defined in the CMAH, forecasters shared numerous additional considerations that affect their avalanche problem decisions. These considerations revealed many possible explanations for the observed differences between the described problem assessments. Varying risk communication tactics among forecasters was one of the main themes that emerged, with potential implications for how avalanche problems are assessed:

• One area of misalignment we observed involved practices for grouping certain problem types together. One example was using a storm slab problem to describe simultaneous wind slab and dry loose problems.
• Some forecasters sought to avoid the simultaneous forecasting of specific problems, like storm slabs and wind slabs.
• There were discrepancies in the way problems were communicated as they progressed, such as whether a deep persistent problem should be directly forecast or should transition from a persistent problem.
• We observed practices around the relationship between other bulletin messages, such as using a lower threshold to forecast a wind slab problem when there are no other problems, or when a problem is typical to the baseline conditions for the region. For example, a desire for the bulletin to look substantially different when unusual conditions exist might lead to different thresholds for forecasting deep persistent problems in continental versus transitional snow climates.
• Differing assumptions about bulletin users also emerged as a factor for how avalanche problems were assessed. Some forecasters were concerned about maintaining credibility with more experienced or local users, while others focused on users with less experience or from a different region.

Our analysis also highlighted that approaches for dealing with uncertainty were a possible source of inconsistencies in avalanche problem assessments. These included tactics like unlisting a problem but continuing to discuss it elsewhere in the bulletin or on an external platform such as social media. Under uncertain conditions, forecasters commonly recounted an increasing value placed on information coming from peers. Forecast agencies in the Rockies, for example, described rules around building consensus with a certain number of forecasters for a minimum time period before removing deep persistent problems from the bulletin. Forecasters based in the Columbia Mountains, who encounter deep persistent problems less frequently, also mentioned a higher value on peers’ opinions, but did not describe more formalized consensus rules. Finally, the role of personal experience in dealing with uncertainty was expressed. This can relate to the increased confidence to more readily make changes that can come with more years of forecasting, or variations in risk tolerance in relation to specific personal experiences.

System constraints emerged as another common theme influencing avalanche problem selection. Possible sources of differences included:

• the fundamental differences between the available information sources in office-based and field-based forecasting programs;
• software constraints, such as the limit of three problems that can be forecast, which compelled forecasters to group and prioritize problem types; and
• influences around the time when forecasts were issued.

OPPORTUNITIES TO INCREASE CONSISTENCY

Our results highlight substantial variability in the observation types, observation values, and additional considerations forecasters use when adding, removing, and transitioning avalanche problems; but this variability does not necessarily suggest that forecasters fundamentally disagree about what constitutes these avalanche problem types. Instead, it demonstrates the complexities of assessing and communicating avalanche problems. While some of the observed differences are justified, developing more consistent practices for cases where avalanche problems do not fit neatly into their defined boxes is critical in the public forecasting context. This is because consistency is one of the key characteristics of effective risk communication, and recreationists are unlikely to recognize that information presented in a similar format could have a different meaning between forecasters, forecast agencies, and regions.

With the current absence of industry-wide standardization or training specific to public forecasting, individual forecasters and forecast agencies have developed their own risk communication practices. This is reflected in the discrepancies observed in the largely unacknowledged additional considerations. The development of transparent guidance on more general topics could be an important step towards addressing the disparate perspectives observed related to questions, such as:

• Is the primary objective when selecting avalanche problems to provide the most accurate reflection of the hazard conditions or should risk communication objectives be weighted more heavily if they conflict?
• What constitutes an avalanche problem that should be listed in the public forecast? Is a dormant problem still a problem?
• Should avalanche problems in a region be relative to baseline conditions within the region, or consistent across all regions?
• Should avalanche problem inclusion criteria shift over time within the same region, for example with respect to the existence or lack of other problems?
• What criteria should be used to order avalanche problems in the bulletin?
• Which bulletin users should be targeted when assessing avalanche problems, and which communication tactics are most appropriate to meet their needs?

The creation of more detailed decision aids for scenarios around adding and removing specific problems is another attractive approach for fostering consistency (e.g., CAIC, 2022). Decision aids such as simple checklists, flow charts, or more complex algorithms that leverage additional data sources such as numeric snowpack models could be used. Decision aids could highlight observation types, observation values, and additional considerations that should be assessed for a specific problem scenario. Transparent decision aids can also be used as training tools for novice forecasters, and support forecasters moving between regions.

A wide variety of assumptions about bulletin users were expressed by forecasters, in some cases leading to conflicting perspectives about how problems should be listed and in what order. Additional research on bulletin users that creates a shared understanding of the target user of avalanche problem information could provide important background information for developing evidence-based guidance on effective avalanche problem use.

CLOSING

Our study revealed considerable variability in the way public forecasters apply avalanche problems in Canada. We hope our results contribute towards developing shared guidance and decision aids for more consistency in avalanche problem assessments; and that they highlight the value of research on how bulletin users apply avalanche problems in their risk management decisions.

While this study focused on public forecasting and considerations related to public risk communication that are not applicable to other forecasting contexts, the important role of largely unrecognized additional considerations in the application of the CMAH stands out. In other contexts, practitioners may consider whether their own operational practices influence their application of avalanche problems; and, if those practices differ between organizations or applications, how that might affect communication of problem information across professional channels and platforms like InfoEx.

If you are interested in having a closer look at Heather’s research, you can find her thesis on the SARP website.

REFERENCES

Colorado Avalanche Information Center. (2022). Avalanche problem guidance for backcountry forecasters. In CAIC Employee Manual (pp. 82–83).

Statham, G., Haegeli, P., Greene, E., Birkeland, K., Israelson, C., Tremper, B., Stethem, C., McMahon, B., White, B., & Kelly, J. (2018). A conceptual model of avalanche hazard. Natural Hazards, 90(2), 663–691. https://doi.org/10.1007/s11069-017-3070-5

Towell, M. J. (2019). Linking avalanche problem types to modelled weather and snowpack conditions: A pilot study in Glacier National Park, British Columbia [Master’s thesis, Simon Fraser University]. http://www.avalancheresearch.ca/pubs/2019_towell_avprobmodel/