ON 23 January 1975 two slope-parallel deformation gages (Reference SommerfeldSommerfeld, [1975]) were installed in the Cliffs avalanche path at Berthoud Pass, Colorado, just prior to a large storm. A 24 h period was allowed for the gages to bond to the snow. They were connected to a recorder on 24 January, about 19.00 U.T. (zone time + 7 h). On 25 January about 02.30 U.T., a natural avalanche released in the area. It tore out the down-slope gage (H-6) but left the up-slope gage (H-7) in place. Figure 1 shows a highly schematic representation of the snow stratigraphy determined at the crown, about 19.00 U.T. 26 January, and the approximate gage location.
Figure 2 shows deformation records from these two gages. The positive γ-axis indicates gage extension. Because of the recording techniques, there are 10 s running means of the deformations. The deformation rate on each gage increases for almost 5 h preceding the avalanche. After the avalanche the rate on H-7 decreases slightly. Also note that the strain-rate on H-6 averages 2.3 × 10−4 h−1 and does not exceed 3.9 × 10−4 h−1 (the strain-rate just prior to failure) at any time.
On 12 February 1975 another gage (H–4) was placed vertically above the previous position of H-6 on about 40 cm of additional snow. A storm during the period 20–21 February resulted in the deformation shown in Figure 3. There was no avalanche during the second storm. The average strain-rate was 2.8 × 10−4 h−1 but at no time was there an increasing strain-rate.
It is impossible to draw any sweeping conclusions from one pair of observations. However, it has been shown in the laboratory that snow exhibits accelerating strain before failure (personal communication from W. St. Lawrence). Figure 2 shows that snow in the field also exhibits accelerating strain prior to failure, at least under some circumstances. If this phenomenon is typical of snow it could form the basis for a short-term avalanche warning system.
Another interesting point is that the strain, as measured by the 3 m long gage, never exceeded 3.9× 10−4 h−1. R. L. Brown (personal communication) found that a strain-rate of at least 10−2 h−1 was necessary to achieve snow failure in tension: 1.5 orders of magnitude higher than measured by H−6. Therefore, the strain leading to failure must have been concentrated over a much shorter distance than that sampled by the strain gage. This observation supports the idea that snow fails at flaws which concentrate the stress and the strain.