Why Snow Level Forecasts Matter for Avalanche Predictions—and How Your Observations Help
January 2025 - Anne Heggli
Predicting avalanches is a challenging task with high stakes.
Avalanche forecasting is complex and demanding. Each storm's impact on a mountain range plays a critical role in how the snowpack develops, directly influencing avalanche risks. Forecasters must consider a multitude of factors, such as the density of new snowfall, wind conditions during the storm, the state of the snow surface before the storm, and whether the storm transitioned between rain and snow. This complex task is further compounded by the diverse characteristics of mountain ranges—varying elevations, slopes, and angles—all of which affect sun and wind exposure. Adding to the challenge, many areas lack direct observational data. With lives potentially at stake, avalanche forecasting is not for the faint of heart; it requires meticulous attention to detail and unwavering dedication.
Why do snow levels matter?
If you’ve been an observer with Mountain Rain or Snow for a while, you may have heard us refer to “snow levels”, or the elevation at which rain turns to snow for a particular storm. Monitoring the snow level is one important consideration for avalanche forecasting. Rainfall on the snowpack increases the risk of wet avalanches. The added water weakens the snowpack’s structural integrity, making it more prone to sliding. Another concern is when snow levels rise throughout the duration of a storm: rising snow levels can signal an increase in snowfall density. This creates an “upside down” snowpack where heavier, wetter snow accumulates on top of lighter, less cohesive snow. This type of snowfall accumulation increases the danger for storm slab avalanches.
Fortunately, you have been diligently collecting valuable rain or snow data, and the Mountain Rain or Snow team has been working to connect that data with avalanche forecasters. To evaluate the effectiveness of these efforts, we designed a case study in collaboration with the Sierra Avalanche Center and the National Weather Service (NWS) in Reno to see how new snow level forecast tools developed by the NWS do compared to the Mountain Rain or Snow observations collected in the area.
Four unique storm events: Avalanche insights
The study focuses on the Lake Tahoe area within the Sierra Nevada, renowned for its complex winter climate and frequent rain-on-snow events. Covering elevations between 5,250 ft and 10,830 ft (1,600 and 3,300 m), this region experiences diverse precipitation patterns influenced by its geography. On the windward western slopes, precipitation is typically greater, while the lee side eastern slopes are drier due to rain shadow effects. We hone our focus on the 2023-2024 winter with over 7,100 Mountain Rain or Snow observations in the region and 111 avalanches reported from the Sierra Avalanche Center archives. We identified periods of increased avalanche activity and then worked in collaboration with the NWS in Reno to compare the Mountain Rain or Snow observations to the National Blend of Models probabilistic snow level forecasts during four case studies. Each plot shows the deterministic snow level forecast as a solid line and the 10th and 90th percentile snow level range in shading. If you are curious to learn more about the power of probabilistic forecasts, you can watch these videos.
Figure 1. A map of the region in which this research was focused (eastern California and western Nevada). The letters A and B on the map mark the locations of Donner Pass and Echo Peak, respectively, which were the two locations for which the snow level forecasts were studied.
Event 1: Rising snow levels (January 12-14, 2024)
This "upside-down storm" began with snow and transitioned to rain, destabilizing the snowpack and triggering 17 avalanches. Mountain Rain or Snow observations captured rain as high as 6,486 ft (1,977 m) and a mixed phase up to 7,034 ft (2,144 m). Probabilistic forecasts successfully reflected these transitions, aiding forecasters in assessing the elevated avalanche risk due to rapid snow density changes.
Figure 2. Panel (a.) shows the snow level forecast data and Mountain Rain or Snow observations over time for this storm from mid-January, and panel (b.) shows a map of where those Mountain Rain or Snow observations were located. The gray shaded areas in panel (a.) show the 10th to 90th percentile ranges for the snow level forecast at the two studied locations, Donner Pass and Echo Peak, and the solid lines show the deterministic snow level forecast at these two locations. The success of the probabilistic forecast model for this storm is demonstrated by the fact that nearly all of the observations of mixed precipitation occurred within the 10th-90th percentile ranges, very few observations of snow occurred below the ranges, and no rain was observed above the ranges.
Event 2: Cold air trapping (February 3-5, 2024)
Another upside-down storm saw a mix of precipitation types due to cold air damming in valleys east of the Sierra crest. Despite rising snow levels in forecasts, Mountain Rain or Snow data showed snow persisting at lower elevations. This discrepancy highlighted the role of terrain-driven weather patterns, which can complicate forecast validation.
Figure 3. Panel (a.) shows the snow level forecast data and Mountain Rain or Snow observations over time for this storm from early February, and panel (b.) shows a map of where those Mountain Rain or Snow observations were located. On February 4, observers continued to report snow around 1500 meters even though the forecast predicted that the snow level would be around 2000 meters. The observations of snow around 1500 meters, which were mostly from observations in the valleys to the east of the mountains, suggests that complex terrain-related atmospheric factors resulted in snow falling at lower elevations even while rain and mixed fell at higher elevations.
Event 3: Consistent snow levels (February 17-20, 2024)
During this storm, snow levels hovered around 6560 ft (2,000 m), with six avalanches reported. Mountain Rain or Snow observations aligned closely with probabilistic forecasts, demonstrating their utility in real-time validation. The coupling of probabilistic forecasts with Mountain Rain or Snow data provided forecasters with a clearer picture of snowpack stability.
Figure 4. Panel (a.) shows the snow level forecast data and Mountain Rain or Snow observations over time for this storm from mid-February, and panel (b.) shows a map of where those Mountain Rain or Snow observations were located. Since very few observations of mixed precipitation occurred outside of the 10th-90th percentile bands, and observations of snow were nearly always within the percentile bands or above, the probabilistic forecast and Mountain Rain or Snow observations were in fairly close agreement for this storm.
Event 4: Classic Sierra snowstorm (February 29-March 3, 2024)
This event not only featured dropping snow levels and significant snowfall, but it also shows the power of the Mountain Rain or Snow observers when we have many eyes on the sky. A whopping 809 observations were submitted in a 48-hour period! The observations revealed variability in precipitation phase across the region, influenced by topography. Observations east of the crest often showed higher snow levels than predicted, emphasizing the importance of localized data for improving forecast accuracy.
Figure 5. Panel (a.) shows the snow level forecast data and Mountain Rain or Snow observations over time for this storm from mid-February, and panel (b.) shows a map of where those Mountain Rain or Snow observations were located. The variability in precipitation phase by location in this storm is demonstrated by the close overlap of observations of rain, mixed, and snow, as well as the appearance of several observations of mixed precipitation far below the forecast snow level.
Practical applications for avalanche forecasting
Mountain Rain or Snow observations offer actionable insights for forecasters:
Rain-on-snow: Suggests potential weak layers due to density changes or crust formation, increasing hazard.
Rising snow levels: Indicate increasing snow density and potential instability. Real-time data can help forecasters monitor these changes during storms.
Precipitation phase at key elevations: If snow falls on the known weak layers (e.g., surface hoar) then the avalanche risk can persist. However, if a little rain falls on surface hoar, it can destroy the weak layer and decrease avalanche danger. Of course, too much rain causes other concerns.
Why this matters: A collaborative approach to avalanche safety
By combining real-time Mountain Rain or Snow data with nuanced forecasts from the NWS, avalanche forecasters have another tool in their data toolbox when assessing the avalanche danger in their region. This collaborative approach not only improves safety for those in mountainous regions but also demonstrates the power of community in tackling complex environmental challenges. If you wish to read more, you can read this paper that was presented at the International Snow Science Workshop in Tromsø, Norway.
