On contributory factors in avalanche hazard evaluation1

United States Forest Service, Alta Avalanche Study Center, Alta, Utah
Received February 11, 1970

(**We apologize to the author and readers of this paper for the image quality.)

Using 20 years of storm and ramsonde profile data, the probability of avalanche hazard to the Alta Highway and Village is plotted as a function of ten contributory factors. Results indicate that the hazard probability is related strongly to precipitation and wind direction. Certain factors previously thought to be important (wind speed, settlement, and ramsonde profiles) do not seem important for the south facing slopes and large avalanches tested in this study.


Atwater (1954) proposed ten contributory factors for avalanche hazard evaluation. Table 1, constructed from early investigations, shows the original contributory factors with corresponding quantitative and qualitative criteria favorable for avalanching at Alta, Utah, and Loveland-Berthod, Colorado. There are two important reasons for further exploration of contributory factors:    

(1) They are conceptually simple. This is important in a field keyed to subjective interpretation of data which contain relatively large amounts of scatter.
(2) They can be studied by means of nearly 20 years of data from Alta and other areas. Pedagogy  would suffer if we excluded this backlog of information.



Analysis of Storm Reports and Snow Profiles

In order to test and draw conclusions con concerning a set of contributory factors we have analyzed 107 storm reports and 87 ram profiles prepared by the Alta Snow Rangers during the period l950 to 1969. The seasonal interval of investigation has been selected as 15 December to 31 March, which represents a compromise of narrowing the interval without seriously depleting the data. Atwater suggested that solar radiation be considered as an additional factor. There have been no continual records made of this factor at Alta so we have removed the predominance of radiation effects by cutting all April storms from this investigation . Available data did not permit analysis of all the original contributory factors. Instead we have analyzed the following set which is supported by data and hence reflects those factors considered important by personnel stationed at Alta:

(1) Total precipitation                                                                                                                                 
(2) Average precipitation intensity
(3) Maximum precipitation intensity.
(4) Gross amount of new snow.
(5) Settlement of new snow.
(6) Wind speed.
(7) Wind direction.
(8) Maximum temperature change during any storm interval.
(9) The precipitation intensity factor devised by Atwater which we designate the "Atwater Number".
(10) A "ram ratio" (snow load/ram strength)                                                                                             .

Factors 1 through 9 are meteorological observations for an arbitrary and not clearly defined period of precipitation, loosely referred to as a "storm" by avalanche personnel.  Factor 10 is the structural observation of the snowpack for the same period.

First, scatter diagrams were prepared of the actual hazard that developed on slopes of common aspect versus each of the factors.  Figure 1 shows the scatter diagram for the factor, total precipitation.  We have elected to work with slopes that pose a serious problem to the Alta Highway and Village, the analysis could be repeated  for slopes that threaten the ski area. Because of their southern aspect these slopes do not form extensive layers of depth hoar which tends to dominate the evaluation problem. A numerical value of the hazard is obtained by summing the magnitudes (see U.S. Forest Service vice Handbook 1961) of slab avalanches that ran during and immediately after the storm period. All slab avalanches, natural and artificial , of magnitudes 3, 4, and 5 are summed. The inclusion of artificial releases can be thought of as a margin of safety. Magnitudes 1 and 2 are omitted since these do not reach the highway or village.

Next, hazard probability diagrams were prepared for each factor (Figs. 2 to 11). Each scatter diagram was divided into 4 or 5 approximately equal intervals on the abscissa. Two probabilities were calculated for each interval: the probability of any hazard of magnitude  3 or greater, P3, plotted as solid lines; and the probability of a hazard of 10 or greater, P10, plotted as dashed lines. More specifically, from an interval on a scatter diagram



From Figs. 2 to 11, we note that the shape of P3 can fall into one of two groups: either P3 can show a general trend to increase as the factor  increases, or P3 can oscillate about 0.5. We designate the former shape as "first order" and the latter as "second order". The same distinction  can be made for P10 except the oscillation is about 0.2 for "second order".
Thus, we observe the following:

(1) With respect to P3, the factors with first order effects are 1, 2, 3, 4, 7, and 9; and with second order effects 5, 6, 8, and 10.
(2) With respect to P10, all factors with the exception of 5 and 6 have first order effects.
(3) Both P3 and P10 are more sensitive to maximum precipitation intensity than to average precipitation intensity.
(4) Wind direction is a better factor than wind speed.
(5) For P3, the WNW maximum may reflect competition between lee effects and orographic effects.  The Wasatch Range is oriented north-south.
(6) For P10, a rising temperature during the storm indicates an increase in the hazard.

The "Atwater Number" (Fig. 10) has a numerical value equal to the amount of water, expressed in cm, which has been deposited for a continuous period at a rate of 0.25 cm h-1. For P3, the Atwater Number has too flat characteristic to be especially useful. However for P10, a transition from 3.8 cm to 6.3 cm seems to be a meaningful indicator of high hazard. It may be possible to base the Atwater Number on wind direction rather than wind speed. Also, if the lower limit on precipitation intensity is reduced to 0.125 cm h-1, then the Atwater Number may have more sensitivity for southern slopes at Alta.

Figure 11 is based on ram profiles taken at the Alta Study Plot2 during the period 1954 to 1969. For each profile on record, a maximum value of the "ram ratio" was determined by trial and error.  The ram ratio corresponding to a particular storm is found by linear interpolation of the profiles taken before and after the storm. For P10 this factor has only a small first order effect when applied to slopes and profiles of southern aspect. The situation may be quite different on slopes of northern aspect.


The results of this study on large avalanches which release on the south facing slopes above the road and village at Alta, Utah, indicate that the probability of an avalanche hazard varies considerably with precipitation and wind direction, only slightly with temperature change, and seems to have no definite relationship to wind speed and snow settlement. The strong dependence of the hazard probability on the maximum precipitation intensity of the storm suggests that the alpine snowpack is sensitive to the rate of stress application. It appears that the Atwater Number, which expresses the combined effects of precipitation intensity, quantity of precipitation, and wind, should be modified to account for wind direction instead of wind speed. The hazard probability and the ratio, snow load/rarn strength, have only a small relationship.

It would be interesting to see how the foregoing compares with results from other areas, not so much as to specific values, which we would expect to differ, but with regard to shapes of curves and whether or not a factor is important.

1 Presented at the Journal, 7, 414 (1970)
2 The Alta Study Plot, located on a flat portion of the valley (elevation m) only partially describes conditions in the fracture zone of southern exposed slopes (elevation 3700 m).



This study was undertaken by the A1ta Avalanche Study Center, Wasatch National Forest, as part of the administrative studies program for avalanche forecasting and control. Guidance given to the author by Dr. E. R. LaChapelle  is gratefully acknowledged. The author is also indebted to Snow Ranger Will Bassett for this thorough collection of storm records during the period 1967 to 1969.

ATWATER, M. M. 1954. Snow avalanches. Scientific American, 190, No. 1, pp. 26-31.
BORLAND, W. M. 1954. Investigation of snow conditions causing avalanches. Interim Report No. 3, U.S. Army Corps of Engineers, Denver, Colorado.
U.S. FOREST SERVICE 1953. Avalanche handbook. United States Department of Agriculture, Washington, D.C. Out of print.
_________ 1961. Snow avalanches: A handbook of forecasting and control measures, United States Department of Agriculture, Agriculture Handbook No. 194, Washington, D.C.


Fig 1. Avalanche hazard to the Alta, Utah, road and village versus total precipitation of storm period

Fig 2.  Hazard probability versus total precipitation. P3, solid line. P10, dashed line.

Fig. 3. Hazard probability versus average precipitation intensity.

Fig. 4. Hazard probability versus maximum precipitation intensity.

Fig. 5. Hazard probability versus gross new snow.

Fig. 6. Hazard probability versus new snow settlement.

Fig. 7. Hazard probability versus average wind speed.

Fig. 8. Hazard probability versus average wind direction.

Fig. 9. Hazard probability versus maximum temperature change.

Fig. 10. Hazard probability versus the Atwater Number.

Fig. 11. Hazard probability versus the ram ratio.

                                                         LEGAL STUFF
Copyright 1998-2001 WestWide Avalanche Network
All Rights Reserved.

Questions or comments to: webmaster@avalanche.org
Last changed: July 11, 2002