|
|
CreepandFailureofAlpineSnow:MeasurementsandObservations
H.Conway1,S.Breyfogle2,J.B.Johnson3andC.Wilbour2
1University of Washington, Geophysics, Box 351650, Seattle, WA 98195
2Washington State Department of Transportation, Box 1008, Snoqualmie Pass, WA 98068
3US Army Cold RegionsResearch and Engineering Laboratory,Ft Wainwright, AK 99703
|
|
|
|
Key Words-rain-on-snow, creep, avalanches
ABSTRACT
We investigated the creep behaviorof alpine snow in an
effort to helpunderstand and predict the timing ofava-
lancherelease.Measurementsofmotionofglideshoes
buried within a natural snowpackshow strains within low
densitysnow are typicallylarge (often exceeding 70%).
The rate ofdeformation increases withtemperature and
is especially rapid in the presence of liquid water.Creep
rates decrease rapidly asthe snow densifies.The slope-
parallelshearingcomponentofmotionismuch smaller
than expectedfrom the usual constitutive assumptions for
snow. Even when snow is first wetted and on slopes up to
36º,the resultant direction of motion is typicallyclose to
vertical.We explain this apparently anomalous behavior
by considering the effects of metamorphic processes and
"capillarystrain"(when liquidwaterispresent)which
causedeformationindependentlyofgravity.Itiswell
knownthatavalancheactivityusuallyincreasesatthe
onset of rain,long before liquid waterhas penetrated to
depth. Wediscuss how capillary induced shrinkage at the
surface might alterthe distribution of stress through the
slab sufficiently to cause existing zones of deficit (or"su-
perweak spots") to extend in length. A rain induced sur-
face alteration occurs rapidly overa wide region and has
the potential to perturb all existing zones of deficit simul-
taneously, thereby increasing the possibility of slope fail-
ure.The analysispredicts slope failure ismore likely if
the overlying slab is thin and the stability is already close
to critical.Field observations ofbehavioratthe onset of
rain support this prediction.
BACKGROUND
Insitumeasurements of creep on slopeshave been used
by others to establish constitutive laws forsnow. Consti-
tutive laws are needed to describe stress-strain relation-
ships and to formulate models forpredicting snow slope
stability. Past measurementshave been madeby monitor-
ing the motion oflight-weight tracers such as ping pong
balls(Perla,1971)orchanges intiltofpolesplacedin
snow(McClung,1974).Mostmeasurementshavebeen
madein well settled snow, and here we presenthigh reso-
lution measurements ofdeformation withinlow density
snow. We are particularly interested in the effects of wet-
ting - some measurementsweremade duringrain-on-snow.
Models of dry snow slope failure suggestthat two condi-
tions must be satisfied before avalancheswill release: the
downslopecomponent ofthe weight ofthe slabmustbe
close to the average shear strength of a buried weak layer;
the rate of deformation within the buried weak layer must
be sufficiently high to causefailure (McClung, 1979,1981;
Bader and Salm, 1990).McClung (1981) analyzed two ex-
treme types of avalanche release mechanisms: (I)- where
increasingstressescausedbyloadingexceedthepeak
|
shear strengthof a basal weak layer over some critical area;
(II) - where conditionssomehow become favorable to cause
an existing flaw in the basal layer to propagatewithout ad-
ditional loading. McClung pointed out that inreality, the
likely mechanism of release for most dry snow avalanches
probably lies betweenthese two extreme scenarios.
Observations indicate that most natural dry snow slab
avalanches release during storms as a result of rapid load-
ing fromsnowfall (McClung and Schaerer,1993).These
avalanches are thought to be examples of type I behavior.
Observations also indicate that avalanche activity usually
increaseswithinafew minutesoftheonsetofrain-on-
snow. These avalanches typically release as slabs several
hoursbeforeliquidwaterhaspenetratedtothesliding
layer; theshear failure atthe basal layer is within dry snow
(ConwayandRaymond,1993;Heywood,1988;Conway
etal.,1988).The release mechanism isclearlydifferent
fromthe more commonly reported scenario forwet slab
release where it is thought that infiltrating waterhas the
effectofincreasingthe stressandweakening a sub-sur-
face layer (eg.McClung and Schaerer, 1993).Further,the
increaseingravitationalloadingfromtheadditional
weight of rain is usuallysmall at the time of release;the
type I contribution issmall.
The mechanism of release of these immediate type ava-
lanches has been a puzzle; it is not clear how the snow at
theshearplaneatdepth knowsthatitisrainingatthe
surface.Froman operational perspective it isimportant
to understand that these avalanches release much sooner
thanwouldbeexpectedifloadingand/orlubrication
causedthe failure.Inprinciple itshouldbepossibleto
predict the time of avalanching to within a few minutes
by predicting the timing of transition fromsnow torain.
In practice,forecasting meteorological conditionsin the
start zones of avalanches is not always straightforward.
Inthispaperwepresentand discussobservationsof
snow stabilityand measurements of snow creep. Wedis-
cuss the observed macro-scale behavior in context of grain-
scale processes.
EXPERIMENTALMETHODS
Observationsofweather,avalancheactivity,snow
stratigraphy and deformation were made in the Cascade
mountainsnearSnoqualmiePass,Washington. Theter-
rain near Snoqualmie Pass lies between 900mand 1700
mand mid-winterrain is common at these elevations.A
typical snowpack in the regioncontains a relatively ho-
mogeneous base 2 to 3 m deep that has settled and grain-
coarsenedduringoneormoreepisodesofrain.Storms
typically deposit up to 1 m or more new snow and subse-
quent rain often causes some ormost of the new snow to
avalanche.
Figure 1 showsthe experimental setup used to meas-
uresnowdeformation profileonslopes.Measurements
were made near the middle of a 200 m long slope inclined
|
75
|
|
|
1
2
3
4
5
6
