M o u n t a i nWe a t h e ra n dS n o w p a c k

Snow densities ofthe buried surface hoar and the layers
immediatelyabove and below show a trend similar to but
more pronounced than the time series fromthe previous
winter.Figure 6a shows that the densitiesof the surface
hoarand the snowaboveand below,within 2cm,were
quite similaron 5 January. Again the adjacent snow lay-
ersdensifiedatmorerapidratesthanthesurfacehoar.
Measurements of the surface-areato volume ratio provide
evidencethat grain growth accompaniedthe densification,
asshowninFigure6b,whiletheburiedsurfacehoar
showed little change.

GENERALOBSERVATIONSANDDISCUSSION

Layers of buried surface hoar have microstructure that ex-
hibitsextreme susceptibility to shearfailure.The geom-
etryand spacing ofthecrystalsasobservedinthefield
and on sections provide strong clues to the reasons.Sur-
face hoarcrystalsspread out as they grow upward (Lang
et al., 1985;Breyfogle,1987),which can result in an um-
brella-like mechanism whennew snow falls. This depends
on weather eventssubsequentto hoar formation (Breyfogle,
1987).Both observationsin the snow pit and on section
cuts showed large gaps between the surface hoar crystals
in this study.Figure 7a shows a classified section image

Figure 4. Measured shear strength, rutschblock and grain sizes
plotted as time series with calculatedrutschblock score for surface
hoar layer buried 28 December 1995. Indicator shows time before
which avalanches were observed.

than5.5andnoneoccurredwhenthecalculated
rutschblock score was 5.5 orhigher.
The heli-sking operation that surrounds the study site
reported natural and skier-triggered slab avalanches slid-
ing on the December 28th surface hoar on 7-10, 12 and 15
January, but none after 15 January 1996.
Initially we observed surface hoar crystals to 20 mm in
length.Although the following 11 weeks showed round-
ing and smoothing of the grains,12 mmcrystals were re-
ported on most test days including the final day. Analysis
of the digitized photographs supports these observations;
we measured little change in the average dimensions of
thesurfacehoarcr ystals.Thephotographsofthe
disaggregated crystalsdo notshowa decrease in crystal
size but rounding is apparent. The smallest reported crys-
tals,2-6mm,were reported on 6 February in the middle
of the observation period,which couqbe due to spatial
variability.However,thesubjectivenature ofcollecting
and observingcrystals probably plays a role in thereported
variability.
Measurements of the layerthickness fromthe section
planes show the surface hoarsettled by somewhat more
than 35% from 5 January to 18 January 1996, as shown in
Figure5.Afterthattherate ofcompactionslowedsub-
stantially,as with the previous case.While the standard
deviationofthemeasurements remained low,again the
rangeofmeasurements overlappedformostofthetest
period (Figure 5).

Figure 6. Density(a) and surface-to-volume ratio (b) of layer of surface
hoar buried 28 December 1995 (V), and layers above (U) and below
(L).Measurements were made from section cuts.

as an example. The mechanical connectionacross the sur-
face hoar, fromthe layer above to the layerbelow, largely
relies on the bases of the surface hoar crystals.
Throughouttheentiretestperiodinbothcases,the
averageslope-parallel cross section of ice grains (the mean
ice intercept) in the surface hoarlayerwas at least twice
that inthe layers aboveand below.Thismeans that the
individual crystals had higherstrength than grains in ei-
therthe layerabove orbelow. The numberof ice connec-
tionsbetweentheburiedsurface hoarandsnowabove
and below was minimized at the base.Thus the number
of connections perunit area contributestothe very low
shear strength. Observations on section cuts show that this
variabledoesnotchange much overtime,yetthe shear
strength of the layer increases.
Mechanical tests and avalanchereleases show failures
at base of surface hoarlayers in shovel,compression and
rutschblocktests,andinslabavalanches(Schweizeret
al., in preparation). Frequently surface hoar crystals grow
on a snow surface beneath which depth hoar or facet crys-
tals develop because the same meteorological conditions
can causeboth (e.g., Langet al., 1985). Figure 7aalso shows

77