Avalanche Release and Snow Characteristics, San Juan Mountains, Colorado Final Report 1971–1975
INSTAAR Occasional Paper 19
1976, 256 pp. 7 plates. Copy reprinted from scan. (cost: $15)
This final report covers research conducted by the San Juan Avalanche Project, Institute of Arctic and Alpine Research (INSTAAR), University of Colorado for the period August 1971 to June 1975. The research was supported by Contract No. 14-06-D-7155 with the Division of Atmospheric Water Resources Management, U.S. Bureau of Reclamation, Department of the Interior, and has had as its purpose the study of the nature and causes of snow avalanches within the vicinity of Red Mountain Pass, Molas Pass, and Coal Bank Pass in the San Juan Mountains of southwestern Colorado. The ultimate objective of the project was to develop a methodology to accurately forecast avalanche occurrences through study of the complex relationship which exists among terrain, climate, snow stratigraphy, and avalanche formation. When the project was initiated, only a limited amount of climatological data was available for the study area. Recognizing that an avalanche prediction model relies heavily upon data gathered from highly accurate, reliable instruments installed on carefully selected sites, a network of fixed instrumentation was utilized to measure meteorological parameters, determine physical properties within the snowpack, and detect avalanche events.
The primary snow study site located at Red Mountain Pass (3400 m) included instrumentation to measure air temperature, temperatures within the snowpack, wind speed and direction, precipitation rate and amount, snow settlement rate, and net all-wave radiation at the snow surface. In addition an isotopic profiling snow gauge provided snow density and water equivalent values throughout the snowpack at 1.0 cm intervals. Seismic and infrasonic instrumentation for avalanche event detection was investigated during the first two winters, but neither of these systems proved feasible.
Detailed investigations into the physical properties of the snow within the study area were prompted by the fact that the San Juan Mountains exhi bit climatic extremes not found in more northerly latitudes where most practical and scientific knowledge of snow avalanche formation has been accumulated. The combination of high altitude, low latitude and predominately continental climate produces a specific <i>radiation snow climate</i>. Generally, this condition is the result of two factors. First, the extreme nocturnal radiational cooling occurring on all exposures produces snowpack temperature gradients of a magnitude sufficient to cause significant recrystallization or temperature-gradient metamorphism. The second factor is the substantial amount of solar energy available to slopes with a southerly exposure. This daytime condition causes melt at the snow surface and subsequent freeze-thaw crusts. These two situations continue to influence the snowcover throughout the winter. The resulting stratigraphy is highly complex and often unstable.
During the second winter many snow pits were dug to collect data on snow stratigraphy. These snow pits were of three types. One type was located at standard, level snow study sites, while a second was located on test slopes or avalanche release zones. Special emphasis was given to the third type associated with the actual avalanche fracture lines. The first two types are acquired as a series at fixed sites to determine changes in snow structure with time. During the third and fourth winters, these received the major emphasis with particular attention directed towards the temperature gradient process. Snow temperatures were measured throughout the depth of the snowcover on a daily basis at sites at three different elevations. Periodic snowpits at these sites demonstrated the relationship between the magnitude of the temperature-gradient and the type and extent of subsequent metamorphism.
As a part of the daily operational procedure during the 1972-73, 1973-74, and 1974-75 winters this project produced an “in-house” stability evaluation and avalanche occurrence forecast for the research area. Such forecasts were made for each 24 hour period and at more frequent intervals during storms. Each avalanche occurrence forecast was evaluated the following day in terms of actual conditions and events subsequent to the initial forecast. During the third winter the avalanche forecast procedure was further refined to give forecasts for specific groups of paths, as well as general area forecasts. Methods employed by the field observers to evaluate numerous meteorological and snowcover parameters in order to produce an avalanche forecast were isolated and described. Forecasting accuracies of 81 percent for the general area and 73 percent for specific path groups were achieved. On the completion of the third winter’s data collection, work began on the development of a statistical model for the purpose of avalanche prediction.
Following the fourth winter’s research, the statistical forecast model was further refined. During this final winter an unusually high level of avalanche activity prevailed, allowing twice the annual average number of avalanche events to be included in the statistical analysis. The stepwise discriminant function program allowed stratification of avalanche and non-avalanche days in terms of antecedent conditions described by ten variables over five, three, and two-day periods prior to each avalanche or non-avalanche day. Analysis suggests that the two-day time step is most efficient, thus reducing the amount of computation, with no loss in forecasting precision. A clear difference is found between dry snow and wet snow avalanche conditions. The dry snow avalanche days are most clearly identified by reference to precipitation totals during the few hours prior to avalanche release and by air temperature over varying time periods according to the magnitude of event being considered. The wet snow avalanche days are best related to the mean and maximum two hour air temperatures in the 12 to 24 hour period prior to the avalanche event. While rapid temporary warming may often preceed cycles of small wet loose avalanches, a more prolonged period of warming is required for larger wet avalanche cycles to occur. A measure of the relative distance of a discriminant score from the discriminant index allows a more precise forecast than a simple “yes” or “no.” This refinement enables the forecast to be stated in probability terms, an approach not previously attempted in numerical avalanche forecasting.
Evidence suggests that avalanche release within sub-freezing snow layers is primarily dependent on precipitation to trigger unstable layers deep within the snowcover. Delayed-action events are extremely rare. While avalanche frequency and magnitude are influenced by precipitation rates and amounts, they are thus determined primarily by the snow structure which exists within the release zone at the time precipitation-loading occurs. Avalanche magnitude is further affected by mechanical strength of all snow layers in mid-track, for this determines the penetration depth of sliding snow and the ultimate volume of the moving avalanche.
In conclusion, the claim is made that the Silverton Avalanche Research Project has been able to produce for the first time an approach to an operational real-time statistical forecast model. This model which, for major avalanche cycles during the dry and wet snow seasons, has an accuracy of 88% and 82% respectively, is also the first to be applied to groups of starting zones and individual paths, and to predict magnitude of avalanche occurrence.PDF (23 MB)