WATER RESOURCESRESEARCH,VOL. 29, NO. 8, PAGES2485-2490,AUGUST 1993
PermeabilityMeasurementson New and EquitemperatureSnow
R. A. SOMMERFELD
U.S. Departmentof AgricultureForest Service,Fort Collins, Colorado
J. E. ROCCHIO
U.S. Departmentof AgricultureForestService,Sonora,California
Measurementsof the air permeabilitywereperformedon snowfrom six horizonswithin a snowpack
at the GlacierLakes EcosystemsExperimentSite in southeastern
Wyoming.The snow was new or
equitemperature
snow,with densities
(Ps)rangingbetween
0.134and0.367Mg m-3. The specific
surface area of ice in each samplewas measured stereologically.Intrinsic permeabilities (K0)
calculated
fromthesedatarangedfrom3.0 x 10-lø to 3.1 x 10-9 m2. The datafittedthe equation
K0 = 1.096x 10-Be-9-57ps,
witha standard
deviation
of2.8 x 10-mm2. A dimensionally
correct
relationshipbetween permeabilityand specificsurfacearea did not fit our data as well as a simple
densityrelationship.The smallscatterreportedis attributedto carefulselectionof undamagedsamples
and the preventionof samplesublimationduringthe measurements.
INTRODUCTION
Duringthe month of February, between 46 and 53% of the
land area of the northern hemisphere is snow covered
[Barry,1992].Continental snowpacksact as chemicalreservoirs;pollutantscan accumulate in the pack over the entire
winterand are releasedduring a relatively short springmelt
period.Interactionsbetween the snow and the atmosphere
can changethe quantities of different chemical species
storedin the snow. In addition, snow can retain chemical
species
to be incorporated into major ice sheetsand large
glaciers.
The record of atmosphericchemistrypreservedin
thiswayis importantin understanding
the pasthistoryof the
atmosphere
[Dibb et al., 1991; Bales and Dibb, 1992].
Processes
that may be important in the acquisitionand
tion was not observed. Field experiments by Sturm and
Johnson [!991] indicate that extreme thermal gradients are
necessaryfor even intermittent convection.
Considerable scatter in previous measurements of air
permeability have made accurate estimates difficult and
constitutea major source of uncertainty in estimating Rayleigh numbers for snow. Bader [1954] showed a range of
about103for snowof density0.3 Mg m-3. KeeIer's[1969]
data indicate a range of about half of Bader's [ 1954]. Shimizu
[1970]and Martinelli [1971] show a range of about a factor of
four at the same density. Shimizu [1970], whose work seems
the mostcarefuland complete, related the air permeability to
snow density (Ps) and grain size (do) for a limited type of
snowwhich he characterizedas fine-grainedcompact:
K0 = 0.077d•e-7'8p•.
redistribution
of impurities in snowpacksinclude wind
pumping[Clarke et al., 1987; Colbeck, 1989; Clarke and
(2)
Regressionswe performed using his data showed that the
Waddington,
1992;Albert and McGilvary, 1992];convection datafitsweresignificantly
improved
by theinclusion
of d•,
[Akitaya,1974;Palm and Tveitereid, 1979; Klever, 1985; as opposedto a simple density relationship.
Akitaya [1974] failed to observe convection when he used
Powerset al., 1985;Brunet al., !987; Sturm and Johnson
1991];and melt water flow [Colbeckand Davidson, 1973]. (2) to calculate Rayleigh numbers which were above critical
Accurate
quantification
of theseprocesses
requireaccurate for convection in snow. Denoth et al. [1979] estimated
permeabilitiesfrom water flow. His results showed large
Theintrinsicpermeability
(K0) is
scatterbut averagedabout half the permeabilitiespredicted
permeabilityinformation.
by $himizu's [1970] equation from Denoth's densities and
Q = KoA(AP/L tzf)
(1)
grainsizes.Theseresultsseemto indicatethat the perme-
whereQ is the volumetricdischarge(cubicmetersper abilitiesgivenby (2) are higherthan the true permeabilities
•cond),K0 is the intrinsicpermeability(squaremeters),A
for the snows they studied.
The measurement
of permeabilityis prone to systematic
errors
which
tend
to
causespuriouslyhigh permeabilities.
pressure
gradient(pascalsper meter),and • is the dynamic
is thecross-sectional
area (squaremeters),AP/L is the
viscosity
of the fluid (kilograms
per meterper second) Thefirsttypeof erroris physicaldamagecausedby sample
handling.
Thiscanbe in theformof cracksor disaggregation
[Friedman
andSanders,1978].
It is uncertain whether or not convection is common in
snowat normal earth surface conditions. Calculated Ray-
leighnumbers
haveexceeded
thosethoughtcriticalfor
causedby insertionof the sampletube, or separationof the
samplefrom the sampletube wall becauseof settlingor
elevated temperatures.
Sommerfeldand Rocchio [1989] also pointedout that
convection
in snow[Nield, 1968]andin laboratoryandfield
when
differentair permeabilitydata are compared,the
experiments
[Akitaya,1974;Brunet al., 1978]butconvetscatterand the average values'seem to be a function of air
Thispaper
isnotsubject
to U.S. copyfight.
Published
in 1993by flowrate.Theyspeculated
thatdry air mighterodeprefertheAmerican
Geophysical
Union.
entialchannels
in somesamples,
increasing
theexperimental
Paper
number
93WR01071.
errorandcausing
thereportedpermeabilities
to averagetoo
This file was created by scanning the printed publication.
Errors identified by the software have been corrected;
however, some errors may remain.
2485
2486
SOMMERFELD
AND ROCCHIO:PERMEABILITYMEASUREMENTS
ON NEW AND EQUITEMPERATURE
SNOW
high. A third sourceof systematicerror is that erosionwould
TABLE 1. SnowSamples
Reported
in ThisStudy
have a more seriouseffect alongflow pathsinvolving sample
Stereological In situ
damage, exacerbating any damage effects. The fact that
Sample
Density, Density,
$•,,
these sources of error are systematic means that accuracy
No. Type mgm-3 mgm-3 Ko,m2 mm-•k2
may not be improved by larger sample numbers if such
A IIA.1
0.191
0.225 1.50E-09 9.2 2.04
systematic errors are large.
lB IIA.1
0.219
0.217 5.82E-10 9.6 0.97
Shimizu [1970] recognizedtwo types of systematicerrors;
2A IIA. 1
0.179
0.177 1.53E-09 8.1 I.•
sampledamageand erosion. His permeabilityapparatuswas
2B IIA. 1
0.167
0.182 1.93E-09 6.4 1.15
designedto decreasethe effectof damagealong samplewalls
3A nd
0.299
nd
7.43E-10 7.6 1.12
3B nd
0.316
nd
5.35E-10 8.3 1.05
but not damagewithin the body of the sample.He addressed
4A IIA.2
0.367
0.307 3.00E-10 8.6 0.83
the erosion problem by measuringthe permeability of a
4B IIA.2
0.306
0.305 8.40E- 10 6.2 0.87
singlesampleduring a 10-hourperiod. However, this exper7 IB
0.152
0.149 2.62E-09 7.6 2.09
8 IB
0.134
0.146 3.07E-09 8.5 2.85
iment was not conclusivebecause(1) The temperaturewas
10 IB
0.150
0.160 2.62E-09 6.3 1.42
low and the humidity was high in his laboratory when this
11 IB
0.174
0.139 2.57E-09 7.0 1.89
experiment was conducted,reducing the amount of erosion
12 IB
0.136
0.140 3.08E-09 6.8 1.85
that might occur, and (2) the snow he used was relatively
14 IB
0.168
0.130 2.65E-09
6.7 1.74
0.311
0.329 1.08E-09 6.8 1.38
highin density(0.42 Mg m-3) and may not havebeen 2C IIB. 1
sensitiveto channel erosion. The permeabilitymeasuredin
TypefromSommerfeld
andLaChapelle[1970].lB, newsnow•
this experiment was near the low end of his measurements, windblown;IIA.1, equitemperature,
decreasing
grainsize,begingrouped with the majority of his measurements in this ning; IIA.2, equitemperature,
decrevising
grain size, advanced;
increasing
grainsize,beginning;
andnd,not
density range. A scepticalinterpretationof this experiment liB. 1, equitemperature,
determined.
is that when erosion was shown not to be a problem, the
permeability measurementwas low.
Grain size estimates such as used by Shimizu [1970] are
based on subjective judgments [Perla and Sommerfeld, LaChapelle[1970]in an attemptto providea morecomplete
1986]and thus introduceadditionalinaccuraciesin compar- description of the snow.
isonsamong different types of snow. While $himizu [1970]
METHODS
determinedthe cross-sectional
ice area objectively,from the
snow density, the number of grains appearingin a section
Snow samples were collected from six distinct horizons
was estimatedby a more subjectivecountingtechnique:"a within snowpitsdug at the Glacier Lakes watershedin the
complicated-shapedgrain having m remarkable constric- Snowy RangeNational Forest, Wyoming, duringthe winters
tions was countedas m + 1 grains." Comparisonsbetween of 1988-1989 and 1989-1990. In situ snow densities were
objective stereologicalmeasurementsof mean intercept obtained from the pit wall for each horizon at the timeof
lengthsof snow samples[Sommerfeld, !983] and Shimizu's samplecollectionby weightingthe samples.
[1970] diameter estimates show that errors of 30-50% in
A total of 47 snow cores were collected.
Corers of the size
subjectivemeandiameterestimatesare easilypossible.The
subjective nature of the grain size estimatesalso make it
difficult to compare different sets of measurements.On the
we used compressthe snow about 3% radially [Worket al.,
1965] and slow elastic rebound causes the snow to press
against the corer walls after several minutes (F. W. Smith,
other hand, subjectivemeasurementshave been shownto be unpublisheddata, 1980).However, disaggregation
alongthe
consistent.They may be preciseevenif they are inaccurate, corer wall, which can occur with large grained or fragile
with the result that an analysislike $himizu's [1970] can be snow, or settlementduring transportationcan defeatthis
internally consistent.
process. Fifteen of the cores survived handling and trans$himizu's [1970] grain size estimatesshouldbe related to portationto the laboratory.This low rate of samplerecovery
specificsurface area Sv (ice surface area per unit snow (32%) has not been reported by other workers. Damage
volume); a stereological parameter that can be measured included cracks causedby corer insertion, settlementaway
accurately [Underwood, 1970] and relatively easily with from the corerwalls causedby transportation,
and sample
computertechnologyavailablesinceShimizuperformedhis shrinkageaway from the walls causedby above freezing
measurements.K0 can be related to the porosity and the temperatures. Cracks causedby corer insertion were gener-
specificsurfacearea of a porousbed by a Carmen-Kozeny ally not apparentuntil the samplewas removedfromthe
equation [Adamson, 1982]
k2(1- p,o)
3
K0= 8S•2
(3)
corerfor sectionplanepreparation.If the sampleseparated
or macroscopic
crackswerevisibleat thispoint,thesample
wasdiscarded.
Gapsalongthewallweresometimes
detected
by visualexamination.
In addition,samples
whichslideasily
from the corer, indicating lack of wall adhesion, were
wherep•,is thepointdensity(1 - thevoidfraction)andk is discarded.Cracksin the samplesand gapsalongthe corer
the Carmen-Kozeny
constant.For uniformspherical
grains, wallswouldgiveanomalously
highpermeabilities
andexac1/S•andd• arerelated
ina simple
way[Underwood,
1970]. erbateerosionproblems.Two coreseachof four horizons
We presenta new set of permeabilitymeasurements
here were collectedon February24, 1989, six from a single
which are not entirely in agreementwith those of Shimizu
[ 1970].We alsomeasuredthe specificsurfaceareaof eachof
our samplesusinga sectionplane method[Perla, 1982]and
classified each snow sample using Sommerfeld and
horizon on January 17, 1990, and one from a horizonon
January11, 1990.Table1 liststhe snowtype,thestereologicaldensity(pointdensitytimesthedensityof ice),andthe
in situ(gravimetric)densitiesfor the samples.
SOMMERFELD
ANDROCCHIO:
PERMEABILITY
MEASUREMENTS
ONNEWANDEQUITEMPEKATURE
SNOW
COLD
FLOW
METER
•
I
CHRMBER
"
CONDITIONING
COLUMN
SNOW
SRMPLE
PRESSURE
METER
Fig. 1. Schematicof the permeabilityapparatus.
2487
SCCM were averaged.These valueswere alsousedin the
calculationof the Carmen-Kozenyconstant(equation(5)).
Thepermeabilities
ranged
between
a lowof 0.3 x 10-9
andahighof 3.1 x 10-9 m2for snowdensities
between
0.13
Mg/m3 and0.37Mg/m3 (Table1). Thesevaluesare in very
goodagreementwith the lowest of Shimizu's [1970]permeabilities(Figure 2). The seven samplestaken from a single
horizon (numbers 7-14) had permeabilities within about
10%, indicatingexcellentreproducibilityin samplesof the
same
snow.
Snow densitiescalculatedfrom the point density measurements obtained from the stereological analysis compared
well with the in situ density measurements(see Table 1).
Snowdensitywhich is plotted againstpermeabilityin Figure
2 was calculated by multiplying the point density by the
density of ice.
-1
The specificsurfaceareasrangedfrom a high of 9.6 mm
to a low of 6.2 mm-•. The corresponding
permeabilities
were 5.8 x 10-•ø m -2 and 8.4 x 10-•ø m -2 which were not
Thepressure
measurements
for thisstudywereperformed
using
anEquibarType 123differentialcapacitance
pressure the lowest or highest permeabilities. The population vari-
gage
(thesamegageusedby Martinelli).It wascalibrated
to
an accuracy of better than 15½by the Colorado State
University
EngineeringResearchCenter prior to the experiments.The flow meter had a factory specifiedaccuracyof
_+1%,which was verified using a bubble flow meter. The
entireapparatusis shown schematicallyin Figure 1. It is
described
in detail in the works by Sommerfeldand Rocchio
[1989]and Rocchio [1990]. The unique feature of the apparatusis the conditioning column. It was filled with snow
similarto that in the sample column to ensure that the air
flowingthrough the apparatus was at the equilibrium temperatureand humidity for the test sample. This design
preventedthe erosion of preferential air flow channels
throughthe snow sample and assured that sublimation or
anceof the specificsurfaceareaswas 1.05.The K: values
obtained using our specific surface area measurements
rangedfrom 2.8 to 0.8 (see Table 1). Kraus et al. [1953] give
k2 = 25for "randomly
poredmedia."
To provide an estimate of core homogeneity, densities
within the cores were analyzed stereologically. Analysis
showed that 84% of the variance was due to intersample
variance, 5% due to location of the core section within the
sample, and 11% from measurements within each core
section.The conclusionfrom these comparisonsis that each
core was very homogeneous.
DISCUSSION
The snow densitiesin this study ranged from 0.37 to 0.13
deposition
of water vapor did not occurduringpermeability
two samples
of artificialsnowwith
measurements.
We previously determinedthat a flow rate of Mg m-3. In addition,
snow
densities
of
0.548
Mg
m
-3
and
0.616Mg m-3 and
200SCCM(era-3 perminuteat standard
temperature
and
pressure)
with corresponding
filter velocityof 1.6 mm s-1 ,
wasin the laminar flow range [Sommerfeldand Rocchio,
1989].
corresponding
permeabilities
of 3.5 x 10-• m-2 and3.3 x
10-• m-: [Sommerfeld
andRocchio,1989]wereusedin
determiningthe empiricalrelationshipbetweenpermeability
and snowdensit>,(Figure2). The followingrelationshipwas
Themethodfor sectionplane preparationdevelopedby
Per!a[1982]was used for our experiments.For specific obtainedfrom an exponential regressionon the data:
surface
areaandpoint densitymeasurements
withinsingle
4E-009snowcore samples,each core was dividedinto three seqtions
andsurfacesections
werepreparedfor eachdivision.A
totalof45 samplesections
werephotographed
andanalyzed
/it
THIS
STUDY
0
todetermine
specificsurfaceareasandpointdensities.
•
95• CONFIDENCE
5E-009IMAGEPRO[Media Cybernetics,
1989],a commercially
• '•
ONCURVE
FITavailable
softwarepackage,wasusedto convertthe snow
•,,•
SHIMIZU
1970
section
photographs
intothedigitalimagesnecessary
foruse
\ \\xx
xx
x
inthestereological
analysis.The contrast
andbrightness
levels
of thevideocameraweremanually
adjusted
to obtain
thebestdiscrimination
betweenthe ice grainsand the
matrix.Two stereological
parameters,
point densityand
surface
areaper unitvolume,weremeasured
usinga pro-
ß-•
o--.
2E-009
-
E
•
1E--009-
\•\x x
•?
•.\x
XX
x
x
•
x
X •xx
x•
•
.• •x
gramdevelopedby Perla [ 1989].
I:•ESULTS
Intrinsic
permeabilities
K0 werecalculated
using
(1).The
volume
flowrateswerecalculated
fromthemassflowrates,
OE+000-
0.0
o.h
o.,
o.,
0.7
Snow Density Mg/m3
ßeambient
barometric
pressure,
andthetemperature
ofthe Fig. 2. Data from$himizu[1970].Solidcurveis (4), and the
experimentchamber. Three measurementstaken at 200
dashedcurvesindicatethe 95% confidenceintervalon that curve.
2488
SOMMERFELD
ANDROCCHIO:
PERMEABILITY
MEASUREMENTS
ONNEW ANDEQUITEMPERATURE
SNOW
K0-- 1.096x 10- 8e-9'57p'
(4)
The standarddeviationwas +-2.8 x 10-mm 2. This curveis
shownin Figure 2 alongwith its 95% confidenceinterval.
The low scatter in our permeability results tends to
confirmthe validity of the measurementmethodand apparatus design.Excellent reproducibilityis indicatedby the
smallspreadin measurements
from the samehorizon(numbers 7-14, Table 1). The use of a conditioningcolumn
minimizedthe problemof sublimationof preferredchannels
throughthe snow cores by dry air. Care taken to eliminate
damagedsamplesand to selecthomogeneous
snowcontrib-
(X1.0E4-9)
•'3
2
uted to the small scatter in the data.
$himizu[1970]relatedgrainsurface
(d•) to permeability.
""1:
/
!lrl
While exact comparisonbetweenhis determinations
of do
and our specificsurfacearea measurementis not possible,
they both estimate essentially the same parameter and
therefore
should show similar trends.
0.13
Our averagedensitywas0.22Mg m-3 andour average
specific
surface
areawas7.6mm-1. Theequivalent
sphere
diameter for a snow made of uniform sphereswith the same
density and specific surface area is 0.2 mm. This compares
with $himizu's [1970] average of about 0.4 mm for the same
density. As was discussedabove, visual estimatesof .grain
number generally result in lower numbers and thus larger
grain size estimatesthan objective measurementsbecausea
largenumberof very smallgrain sectionstend to be ignored.
We had hoped that by using specificsurfacearea instead
of a derivedequivalentgraindiameter, we would improveon
0.170.210.250.290.33
Snow
Density
(Mg/m^3)
0.37
Fig. 3. Permeability, density, and surface area per gramforthe
samples.
crystalsper volume. The oppositeis true of older snowwith
simple crystal shapes, which have higher densities,less
surfacearea per crystal, and more crystalsper volume.This
effect can also be seen in Figure 3 where snow densityand
surface area per gram are seen to be highly correlated.
Shimizu's
[1970]d• texturalrelationship.
Wehaveonlyone Low-density snow has a high surface area per grambuta
case that can be compared to $himizu [1970]. Our four small number of gramsper volume, while the oppositeistrue
samples,2C, 3A, 3B, and 4B would be classifiedas fine grain of high-densitysnow. The net result is to decreasethe range
compact
snowwithan average
densityof 0.31Mg m-3. A of specificsurfaceareasin equitemperaturesnowof different
regression
onlogK0 versus
1/Svshows
a slope
of 1.8(R2 =
densities.
0.56). This restfitis in general agreementwith the regressions
While specific surface area must have an influenceon
we performed to data taken from Shimizu's [1970] Figure 15 permeability, we conclude that the effect is small for the
for bothfor bothslope(1.22to 2.04)andR 2 (0.32to 0.88). snow we studied because of the density correlationdeThis result is consistent with Shirnizu's [1970] choice of a scribedabove and is maskedby other sourcesof scatterin
squarerelationshipwith grain size which is also dimension- the measurements. These could include differences in tortually correct.
osityandgrainform factorsamongthe samples,undetected
To test the importanceof Sv in our data (without the two inhomogeneities,and undetected sample damage.
artificial samples)we obtained the following relationship:
Our negativeresult with regard to surfacearea callsinto
KoS•= 0.65e-lø'lSø', R2= 0.785
questionShimizu's[1970]relationshipbetweenpermeability
(5) and grainsize. However,becauseof our smallnumberof
$himizu [1970]proposedan approximationto Brinkman's
relationship[Scheidegger, 1960]:
Ko/d•= 0.56e- l•.so,
attentionto a potentialproblem. We note that otherperme-
(6) ability measurementspresentedby Shimizu [!970] are not
This is very similar to (5), providing some theoretical support for our result.
However, for a fit to density only (without the data on
artificial snow), we obtained
K0= 1.36x 10-Be-1ø'97•,
R2= 0.893
samplesandlimitedrangeof specificsurfaceareas,wedo
not think we have a definitivetest, but merelywishto call
consistentwith the data he usedto determine(2). Also,the
use of our lower permeabilities
in calculatingRayleigh
numbers
wouldleadto a prediction
moreconsistent
withthe
resultsof experiments
[Akitaya,1970;Brunetal., 1987].
On
the other hand, our density range of 0.13-0.37 averages
lowerthanthe0.22-0.45Mgm-3 thatShirnizu
(7) significantly
Comparing(5) and(7) showsthat Sv may not be as important
a parameterin determiningthe permeabilityof natural snow
we studiedas we originally thought, at least in our data.
We attribute this result to the small range in specific
surfaceareas amongour samples.The small range among
the snow types we studiedhas a relatively simpleexplanation. Complex crystals associatedwith new snow have more
surfacearea per crystal, lower densities,and therefore fewer
[1970]usedto determine
hisrelationship.
It is possible
that
the typesof snowusedfor the two differentsetsof measurements were not comparable.
It doesnot appearthat the Carmen-Kozeny
theory(equation(5))modelsthepermeability
of naturalsnowsatisfactorily. Our determinations
of k2 rangedover morethana
factorof 3 andwereall considerably
lowerthanthetheoret-
icalvalueof 25.Fornatural
snowthepermeabilities
would
have
tobeabout
afactor
of10higher
toraise
thevalue
ofk2
SOMMERFELD
ANDROCCItlO:
PERMEABILITY
MEASUREMENTS
ONNEWANDEQUITEMPERATURE
SNOW
2489
of snow
to25.Suchhighpermeabilities
wouldbe aboveShimizu'sBader,H., Mineralogicaland structuralcharacterization
[1970]
highest
measurements
for thetypesof snowwe
anditsmetamorphism,
in SnowandIts Metamorphism,
pp. 1-55,
Corpsof Engineers,U.S. Army, Wilmette, II1., 1954.
studied.
Sincethe mostlikely sourcesof errorin these Bales,
R., andJ. Dibb, The GISP2 Ice Core and Snow-Atmosphere
.measurements
aresystematic
andgivespuriously
highvalChemicalExchange,Eos Trans. AGU, 73, 213-215, 1992.
ues,
values
higherthanthosemeasured
by Shimizu[1970] Barry, R. G., Encyclopediaof Earth SystemScience,vol. 1, pp.
517-524, Academic, New York, 1992.
seem
very unlikely.Furthermore,we did not find any
Brun,
relationship
between
thek2 calculated
fromourexperimen-
taldataandthe snowtextureas classifiedin Table 1. While
Shi'mt•u's
[1970]
d02
termissimilar
totheS•2 termin(5),the
exponential
density
termsin his(2)andour(4)givedifferent
relationship
betweendensityand permeabilitythandoesthe
(1- 9)3termin(5).Thusneither
Shimizu's
[1970]
datanor
ourssupports
the Carman-Kozenymodel.
E., F. Touvier, and G. Brunot, Experimental study on
thermalconvection
andgrainspictureanalysis,in Proceedings
of
the NATO Advanced Study Institute on Seasonal Snowcovers:
Physics,Chemistry,Hydrology,editedby H. G. JonesandW. J.
Orville-Thomas,746pp., D. Reidel, Norwell, Mass., 1987.
Clarke, G. K. C., and E. D. Waddington, A three dimensional
theory of wind pumping,J. Glaciol., 37, 89-96, 1992.
Clarke, G. K. C., D. A. Fisher, and E. D. Waddington, Wind
pumping:A potentiallysignificantheat sourcein ice sheets,IASH
Publ., 170, 169-180, 1987.
CONCLUSIONS
Ourresultsshowthat a simplepermeability-snowdensity
relationship
(equation (4)) fits our data better than one
•cludinga dimensionallycorrect specificsurfacearea parameter.Snowdensityis a convenientparameterbecauseof
itseaseof measurement both in the lab and in the field. The
snowtypesand the density range we studied span those
commonly
found in continental snowpacksat the start of
meltandgenerallymake up the bulk of suchsnowpacks.
We chooseto make detailed comparisonsbetween our
resultsand those of Shimizu's [1970] because we think his
arethemostcarefuland completemeasurementsavailablein
the literature.
Colbeck, S.C., Air movement of snow due to wind pumping, J.
Glaciol., 30(120), 209-213, 1989.
Colbeck, S.C., and G. Davidson, Water percolation through
homogeneous
snow, in International Symposiumon the Role of
Snow and Ice in Hydrology, vol. 1, pp. 242-256, UNESCOWMO-IAHS, Geneva, 1973.
Denoth, A., W., Seidenbusch,M. Blumthaler, P. Kirchlechner, W.
Ambach, and S.C. Colbeck, Study of water drainage from
colunmsof snow, CRREL Rep. 79-1, pp. 1-14, U.S. Army Cold
Reg. Res. and Eng. Lab., Hanover, N.H., 1979.
Dibb, J. E., J. L. Jaffrezo,and M. Legrand, Initial findingsof recent
investigationsof air-snow relationshipsin the summit region of
the Greenlandice sheet,J. Atmos. Chem., 14(1-4), 167-180, 1991.
Friedman,G. M., and J. E. Sanders,Principlesof Sedimentology,
John Wiley, New York, 1978.
Keeler, C. M., Some physical propertiesof alpine snow, CRREL
Rep. 271, 67 pp., U.S. Army Cold Reg. Res. and Eng. Lab.,
Hanover, N.H., 1969.
Klever, N., Air and water vapor convectionin snow, Ann. Glaciol.,
However,our results are in disagreementwith those of
Shimizu
[1970]on two importantpoints.Our permeabilityat
6, 39-42, 1985.
anydensityaveragesabouthalf of Shimizu's [1970]average,
andwedidnot findconvincingevidenceof a specificsurface Kraus,G., J. W. Ross,and L. A. Girifalco,Surfaceareaanalysisby
meansof gasflow methods,Steady stateflow in porousmedia, J.
areaeffecton the permeabilitythat would correspondto
Phys. Chem., 57, 330-334, 1953.
Shimizu's
[1970]graindiameterrelationship.
Threedifferent Martinelli, M., Jr., Physicalpropertiesof alpine snowas related to
hypotheses
couldexplainthis discrepancy.First, our data
weatherand avalancheconditions,Pap. RM 64, 35 pp., Rocky
Mt. For. and RangeExp. Stat., U.S. Dep. of Agric. For Serv.,
aretoolimitedin snowtype to showa specificsurfacearea
relationship.
If thisis true, it raisesthe encouraging
possi- Fort Collins, Colo., 1971.
Cybernetics,
Image-ProH, version2.01.00,Silver Spring,
bilitythat snowtypesidentifiablein the field couldbe used Media
Md., 1989.
Mthdensitymeasurements
to makeestimates
of permeabil- Nield, D. A., Onsetof thermohalineconvectionin a porousmeity.Second
Shirnizu's
[1970]datacontainsystematic
errors. dium, Water Resour.Res., 4(3), 553-560, 1968.
Third,ourdataaccidently
contained
onlylow-permeabilityPalm,E., andM. Tveitereid,On heatand massflux throughdry
snow,J. Geophys.Res., 84(C2), 745-749, 1979.
samples.
R., Preparationof sectionplanesin snow specimens,J.
Furtherstudyof the relationship
of permeability
and Pefia,
Glaciol., 28(98), 199-204, 1982.
density
including
a greaterrangeof snowtypesanda greater Pefia,R., Stereology,
Ster. C (version2.2), report,8 pp., modified
rangeof specificsurface areas is indicated. Different sam-
by Glen Brink, May 1989.
plingtechniques
will be necessary
to samplesnowtypes Perla,R., andR. A. Sommerfeld,On the morphology
andsizeof
suchas verynew snowand depthhoarwithoutphysical snow crystals, paper presented at the International Snow Science
ISSW WorkshopCommittee,SouthLake Tahoe,
damage
tothesamples.
Perhaps
suchstudies
willhelpclarify Workshop,
Calif., Oct. 22-25, 1986.
therelationship
between
permeability
andparameters
other Powers,D., S.C. Colbeck,andK., O'Neill, Experiments
in thermal
thandensity
whichshould
existbutappeared
to bemasked convection
in snowspecimens,
Ann. Glaciol.,6, 43-47, 1985.
inourdata.Equation
(5) is verycloseto Shimizu's
[1970] Rocchio,
J. E., Intrinsicpermeability
of naturalsnow,MS thesis,49
approximation
to Brinkman's
[Scheidegger,
1960]theoreti- pp.,Dep.of EarthResour.,Colo.StateUniv., FortCollins,1990.
A. E., ThePhysics
of Flow Through
PorousMedia,
calrelationship
(6).Thisindicates
thata theoryderived
from Scheidegger,
313pp., University of Toronto Press, 1960.
Brinkman's
arguments
mightbeapplicable
to snow
perme- Shimizu,
H., Air permeability
of deposited
snow,Contrib.Inst.Low
ability.
Temp., Sci., Ser. A, 22, 1-32, 1970.
REFERENCES
Sommerfeld,
R. A., A branchgraintheoryof temperature
gradient
metamo.rphism
in snow,J. Geophys.
Res., 88(C2),!484-1494,
1983.
Adamson,
A. W., Physical
Chemistry
of Surfaces,
4thed.,John Sommerfeld,
R. A,, and E. R. LaChapelle,The classification
of
Wiley,NewYork, 1982.
snowmetamorphism,
J. Gldciol., 8, 3-17, 1970.
Akitaya,
E.,Studies
ondepth
hoar,Contrib.
Inst.LowTemp.
Sci., Sommerfeld,
R. A., andJ. E. Rocchio,
TheDarcypermeability
of
Set.A, 26, 1-67,1974.
fine-grained
compact
snow,in Eastern
SnowConference,
ProAlbert,
M.R.,andW.R. McGilvary,
Thermal
effects
duetoairflow ceedings
of the1989AnnualMeeting,
Quebec,
report,pp. 121andvaportransport
in drysnow,J. Glaciol.,38, 273-281,1992.
128,EasternSnowConf., 1989.
2490
SOMMERFELD
AND ROCCHIO:PERMEABILITY
MEASUREMENTS
ON NEW aND EQUITEMPERATURE
SNOW
Sturm, M., J. B. Johnson, Natural convection in the subarctic snow
cover, J. Geophys. Res., 96(B7), 11657-11671, 1991.
J.E. Rocchio,
U.S.Department
ofAgriculture
Forest
Service,
19777 GreenIcy Road, Sonora, CA 95370.
Underwood, E. E., Quantitative $tereology, Addison-Wesley,
R. A. Sommerfeld,
U.S.Department
ofAgriculture
Forest
Set.
Reading, Mass., 1970.
vice, 240 West Prospect, Fort Collins, CO 80526.
Work, R. A., H. J. Stockwell, T. G. Greeman, and R. T. Beaumont,
Accuracyof field snow surveys,WesternUnited Statesincluding
(ReceivedAugust 7, 1992;
Alaska,Tech.Rep. 163,43 pp., U.S. Army ColdReg.Res.and
revised April 12, 1993;
Eng. Lab., Hanover, N.H., 1965.
accepted April 19, 1993.)