SEASONAL SNOW COVER MONITORING USING FMCW RADAR

H. Gubler and P. WeilenmannSwiss Federal Institute for Snow and Avalanche Research

CH7260 Weissfluhjoch/Davos

Stationary, ground-buried and mobile, sledge mounted FMCWradars have been used for five years to investigate snowstratigraphy. The systems have proved their feasibilityfor localized, nondestructive measurements of the developmentof snow layering, waterequivalence, initial percolationof meltwater, flow depth of dense flow avalanches, and slopeperpendicular flow speed profiles. Mobile radars are usedto determine areal distributions of snow stratigraphy andwaterequivalence.

INTRODUCTION

The method of monitoring snowpack stratigraphy,waterequivalence and wetness using FMCW radars hasbeen investigated experimentally and theoreticallyin different papers by Ellerbruch and Boyne (1980),Schmidt, Gubler and Hiller (1984) and Gubler andHiller (1984). The object of this paper is to addrecent applications and results.

Eight FMCW (Frequency Modulated ContinousWave) radars are in use at SFISAR for differentinvestigations: six localized radars buried in theground, looking upward into the snowcover, at fourlocations, and two sledge mounted systems lookingdownward. Two of the radars (X-band and C-band) areinstalled in the study plot of the Institute at2500m a.s.l. These two systems are controlled by acomputer at the Institute. Four systems (X-band)are installed in avalanche tracks at the LukmanierPass to monitor flow heights and slope­perpendicular avalanche speed profiles. The twosledge-mounted systems are used to measure theeffect of trees and topography on snowstratification.

PRINCIPLE OF OPERATION AND DATA ANALYSES

AVALANCHE

REFLECTIVITY

10M

Figure 1.Principle of FMCW radar operation.

antenna to target and back, determines thedifference in frequency between the signaltransmitted, and that received at any instant.This frequency difference df determined by mixingthe actual transmitting frequency with thereceiver's frequency is related to electricaldistance by the equation,

where snowdens and icedens are the densities of

d ~ 1/(1+O.77l(snowdens/icedens» (3)

where c denotes the speed of light. For theSFISAR systems, the conversion factor amounts to

To convert electrical depth 1 to geometricaldepth d, the following relationship provedreliable,

(2)

(1)1 ~ df*P*c/4*B

df/l ~ 2l72Hz/m

The FMCW radar measures electrical distancesbetween the radar and specular reflectinginterfaces within the snowpack, the soil-snow andthe soil-air interface. Electromagnetic radiationfrom the transmitting antenna (Figure 1) is beamedvertically at the snow pack either from above thesnow cover (sledge-mounted systems) or from belowthe ground surface (buried systems). A triangular­modulation oszillator (sweep period P~26.4ms)

sweeps the frequency of the microwave source as alinear function of time between the band limits(3.6 - 8GHz, 8 - l2.4GHz, 13.6 - l8GHz, bandwidthB - 4.4GHz). The second antenna receives thereflected radiation. The electrical distance, from

87

Maroz 188. (Alpl ....1. lfn.)

•

..

10 kHz

, .

.....

I.6•••..-v

~....... ··V &.

.ll2a .488'1

I LA'loA ... .v,..,.S••l.eY

Specular reflection as a function ofelectromagnetic distance (arbitry units)....~l__--.-l, ..II.'U. ~14'

proportion to the inverse of the radar-to-targetdistance squared. The system built at SFISARequalizes radar return signals from targets between0.5 and 7m in medium density snow. Thereforerelative amplitudes within a spectrum will dependon the distance between radar and snow. Thisdependence is clearly shown in Figure 2. Theexample shows that best overall contrast isachieved for a radar to snow distance of about .Sm.The weakly structured middle parts of the spectracorrespond to a heavy storm in January, 1986. Itcan also be seen from this series of spectra thatthe appearent resolution increases with decreasingradar-to-snow distance because of the decreasingfootprint diameter (footprint is the area coveredby the radar beam). Figure 2 shows an enhancedrepresentation of the spectra as described earlier.Only significant peaks are shown. Neverthelessambiguity remains for the weakly structuredmiddle parts of the spectra.

The amplitude of the first sidelobe of the sin(f)/fspectral line envelope is only 13 db down from themain lobe amplitude. As a consequence of this, weakreflections may be masked by nearby stronglyreflecting interfaces. Interferences of strongsidelobes with weak mainlobes originating fromnearby strong and weak reflections may causeambiguities in the interpretation of spectra. Toremedy the problem the spectra should bedeconvoluted using the theoretical line shape, butcomputing time to perform the deconvolution wouldbe enormous. To decrease ambiguity we convolute thespectra with the theoretical envelope of the lobeamplitudes of a single line and subtract theconvoluted spectra from the original one. Peaksremaining in the diffe-rence spectra are consideredto be physically relevant. For a given microwavefrequency, constructive and destructiveinterference between signals of differentinterfaces will occur. Fortunately the occurenceof this type of interference is significantlyreduced by the 1/3 octave microwave frequencysweep. Significant interference occurs only forinterface layers which are separated by less thanone wavelength. A more complete technicaldescription is given in Gubler and Hiller (1984).

LW(-3db) - 0.44*c/4*B (4)LW(-3db) - 1.62cm (35.2Hz)

snow and ice. Frequency difference df is determinedfrom the Fourier spectra of the radar's mixeroutput. For uniform windowing in the FFT (FastFourier Transform) algorithm, the -3db linewidth inunits of distance in air is

For the on- and off-line analyses of the mixeroutput signal, we use a Hewlett Packard series 300workstation with fast A/D and FFT boards. Thecodes allow automatic on-line measurementswith two radars at given times each day, as wellas for off-line high speed analyses from magnetictape. Single spectra may be interpretedinteractively by setting markers to spectral linesand either introducing measured or estimated layerdensities to determine geometrical distances, orintroducing geometrical distances to calculatewaterequivalences. These types of analyses allowfor a direct comparison of radar profiles toconventional ramhardness and morphologicalprofiles.

Codes for automatic spectral line recognitionare used to analyse large time series of spectra.Examples shown in this paper are: time series ofspectra for seasonal snow covers, series showingavalanche flow over a radar, and series visualizingareal variation of the stratigraphy of natural snowcovers.

FIELD TESTS

To investigate system performance under fieldconditions, different types of test measurementshave been performed. Some of these experimentswill be described here, and additional informationis given in Gubler and Hiller (1984).

From the radar equation we may learn that themicrowave return from a plane target decreases in

h.lleY

IA... • ......J............_.-._-...Figure 2. Enhanced spectra measured as a function

of distance between radar and snow. The actualseparation is given in the title of eachspectrum.

88

For almost all spectra shown in this report,~he frequency difference between calculated points(40Hz) is chosen to be about equal to thetheoretical -3db width of a single frequencyspectral line (35Hz). The first line contourminimum is at 80Hz, the first-order side lobemaximum is 120Hz apart from the central lobe peak.Analyses of real spectra at different resolutionshave shown that this is a good compromise betweenincreasing computing time for increasing pointdensity and possible information loss atdecreasing point density. These specificationsalso allow a high time resolution of onespectrum measured every 26ms.

In Figure 3 spectra of two radars looking fromabove into the snowcover and working at differentfrequencies (8-12.4GHz, 13.6-18GHz) are comparedwith the spectra of a third radar also working atthe lower frequency but looking from below the snowcover. The radars were about 15m apart. The mainfeatures of the stratification can be recognized in

.,11 spectra, whereas the fine structure depends onfrequency dependent interferences as well as onlocal variations of the layering and differences inviewing angle and footprint. For the buried radar,which is one of the study plot radars, small cornerreflectors have been mounted at a height of about4m above ground. These reflectors allow anapproximate measurement of local waterequivalencewithout the need of an independent geometrical snowheight measurement. The geometrical distance isgiven by the known height of the reflectors. Forwaterequivalence above 100mm this method producesvery useful results. The disadvantage of thesereflectors can clearly be seen from Figure 3c.Interferences between the strong corner reflectorsignals and strong signals originating fromnear-surface targets led to "mirror" peaks above6.5kHz.

Volume scattering in dry snow for cmwavelengths is very low. The radars measurespecular reflections from interface layers.

18

Figure 4. Series of spectra showing the effect ofnon slope-perpendicular incidence of the radarbeam. a) normal incidence, b-d) incidence 10·from normal at 3 horizontal directions.

L... J

Ie~r.qu.ncy In CkHzJ

Figure 3. Comparision of spectra taken 11 on March1986 from above the snow cover at differentmicrowave frequencies a) 8-12.4GHz, b)13.6-18GHz and c) from below the snow cover atthe lower frequency with "mirroring" caused bycorner reflectors.

;; rCll 1 Harz 1986 (A!ftpl. r.1. 1tn.)

I'"~ a

SeecuJar reflection as a function of•... eectromagnetic distance (arbitry units)::J

~

... .7m .1sev~.

a

8 IiiF'ra.quenz In CklaJ

~ I ! • t 110 kHz.. .7m.Bev

~

!b b

•'0::J

~

1·Ie

. 7m .BevIii

CCklaJ

....>~

• C'0::J Ie~

1 .7m.Bevd

89

Measuring specular reflections with a system havingthe receiving antenna close to the transmittingantenna requires that the radar beam isperpendicular to the reflecting surface. Layerinterfaces in a natural snow cover are neverexactly flat and parallel. The effects ofdeviations of beam directions from normal incidenceby 10° in different horizontal directions are shownin Figure 4. Obviously, snow surface and groundsurface are not in parallel for the case shown. The

finite antenna resolution (7°) permits somedeviation from perpendicular illumunation at thecost of resolution. At L-band frequencies, wheremultioctave antenna having a very low spatialresolution have to be used, depth resolution ispoor. For these cases we use plastic lenses todecrease the footprint and increase depthresolution, with good success.

snowsurface

profile before nearby pitlOS opened

profile after pit was openeddistance beam pitwall 1m

2cm of surface rime removed

1cm soft crust rem oved

2cm soft snow and soft crustremoved

18cm of hardness 2 snowremoved

25cm of hardness 3 snow.3cm of hardness 2 snowand thin crust rem oved

surface roughness increased

8cm hard layer removed

surface roughness increased

21 cm of hardness 3 snowremoved

radar lowered

Hcm of hardness 3 snowremoved

20cm of hardness 3 snowremoved

08cm removed ....

>-radar lowered

....J....::::E

7cm removed >-....J>-0:::

14cm removed <a..z....0:::""

20cm removed <000::::I: ...........:I: 0:::

ground surface >-a.. 0coarse grovel .... z

0<

radar reflection as a function of snow depth (arb.units) electromagnetic profileFigure 5. Selected spectra and series of

electromagnetic profiles of an experimentwhere the snow cover was removed layer bylayer.

90

Comparison Radar Profil - Traditional Prof!!

In the experiment shown in Figure 5, thesnowcover was removed layer by layer. Radarprofiles were taken at each stage. Theelectromagnetic stratifications, determined byusing the code that automatically searches forsignificant spectral lines in consecutive spectra,are plotted together with a selected number ofindividual spectra. Some interesting effects can beseen from this series. Removing the thin very roughand low density surface hoar layer on top of thethin crust made this crust finally visible. Thesurface hoar performed an impedance match betweenthe crust and air, and for this reason decreasedthe reflectivity of the crust. In the originalspectra, not shown here, the hoar-crust combinationcan just be seen. In many cases the interfacesrecognized by the radar are in perfect agreementwith the mechanical interfaces found in the pit. Insome cases of thin crusts, the radar did notresolve the upper and lower boundery of the crust,showing only a weak signal corresponding to one" ofthe bounderies. Interference effects discussedearlier can occasionally mask thin crusts. Effectsof interference and footprint diameter can be seenby comparing the shapes of the spectra for the

Date: 16. Ap r i 1 1985

layers near the ground throughout the experiment.These layers consisted of partially melted andrefrozen coarse-grained depth hoar. The experimentshows that a rough surface has a significantlyreduced reflectivity.

In Figure 6 an electromagnetic profile iscompared to the corresponding ram andmorphological profiles.

Additional experiments have been done in drysnow to check validity of Equation 3 for a widerange of densities, snow stratifications andwaterequivalences. The results support Equation 3within a few percent for shallow snow covers (Theproblem of geometrical measurement is explained inGubler and Hiller, 1984) and within 1% forwaterequivalences above 300mm.

Experiments to measure mean snow wetnessusing a single radar or a combination of an L-bandand an X-band radar were also performed. We foundthat absorbtion, reflection at water-saturatedlayers and basic resolution of the radars limitquantitative measurements severely. But the radarsallow important insights into the verticaldistribution of meltwater during the initial stageof melting of a seasonal snow cover.

SELECTED APPLICATIONS AND RESULTS

Radar - Prof i 1

CD 3o

D.e

II:

We will discuss selected results of fourapplications of FMCW radar at SFISAR: 1) localizedmonitoring of the development of stratigraphy andwaterequivalence during the winter season, 2)damming of water at interface layers during theinitial springtime melt phase, 3) areal variationsof snow stratigraphy, and 4) localized flow depthmeasurements in dense flow avalanches.

MONITORING OF SNOW STRATIGRAPHY

(5)r ~ lfinal/lintit

findens~initdens/(r-(l-r)*(n-l)*initdens/icedens)

Waterequivalence and stratigraphy have beenmonitored in the SFISAR study plot at Weissfluhjochsince 1982 using FMCW radar. A selected number oftypical spectra of winter 1985/86 are shown inFigure 7a. The bitmap presentation of theelectromagnetic profiles is given in Figure 7b. Toimprove this type of presentation at least twomeasurements per day will be made in the future.The settling curves in Figure 7b allow a simplecalculation of the increase of mean densities withtime,

where initdens and findens denote initial and finalmean densities, linit and lfinal, initial and finallayer thickness, and n the index of refraction ofice (1.775). For the example in Figure 7b densityincreased from an estimated 190kg/m3 immediatelyafter the storm to 430kg/m 3 120 days later. Thesenumbers are in good agreement with correspondingfield data. Three selected spectra of Jan. 7/17/20also show the fast increase of snow height followedby significant settling. Often settling can be

Prof i JRamm -

I I I I I ~(~lr r~

11~~Dao(1o;~··.·~- 0 • .. \ II -

F; gure 6. Compad.son-oT-a--rate-w---;"i-n-c"t-e-r-r-a-:d-a-r--- ..Jspectra with the corresponding traditionalsnow and ram profiles. To enhance the spectra,a logarithmic power scale with a lower cutoffat the spectral noise floor has been chosen.Snow pit and radar were about 30m apart.

91

Rederprofl les Study Plot SFISRR

a

Specula,. refl.ctton a* .. function of electromagnetic dlet.nee

Carblta,.y unlta)

13 Sep 1965

l!~

monitored during the storm, as an importantindicator for increasing strength and stabilityBecause these radars can be placed within releasezones without being damaged by avalanches, theyprovide an important tool to monitor snowaccumulation, settling and avalanche releases inpotential fracture zones.

MELTWATER PERCOLATION

Damming of meltwater at crusts or layerinterfaces within the snow cover may lead to thinslush layers of very low strength responsible forwet snow slab releases. Buried X-band FMCW radarspermit monitoring the initial phase of penetrationof surface meltwater into the snow cover, dammingof meltwater, and increasing wetness within layers.Several examples are shown in Figure 8. During theinitial stage of melt of a seasonal snow cover,meltwater seems to percolate along preferedchannels, even in snow below freezing, and is oftendammed by crusts. Increasing liquid content atinterfaces causes a significant jump of the indexof refraction at these dry-wet interfaces. Thisresults in a high reflectivity of the interface,causing a significant peak in the spectrum. If theliquid content within a layer reaches a few percentby volume, microwave absorbtion at X-bandfrequencies increases drastically. Penetrationdepth decreases from several hundred wavelengths indry snow to a few wave lengths in wet snow, maskinganything within or behind the wet layer. Figures8a-e show meltwater damming during spring 83/84/85at the SFISAR study plot and on a nearby northeastfacing slope (Gemsluecke). Indexes for surfacemeltwater production, for the prevailing type oftemperature profile and the amount of waterreaching the ground (measured by a nearbylysimeter) are given for the different years.

l1C..e

'"""'l'I"."_5 _

~1~

~H: I~

~I~~I~l~I~

~I~

I·

.'-~---- -=...J

1 1~7 days

............... --

--'.--........---

...

.'

~

".. :~~. .- .::: ..," .. '.

rlllit

17.January 1986

.: ~ .-=-------'-~--4...;..~...:......:.....:__~_::-:~......;:.. --

'----~----=---=::;~

b

f:

t~i!

f=t 400H.

i'

tt:~

t~.

iU

Figure 7. a) Selected spectra from theSFISAR study plot of winter 85/86 (buriedsystem). b) Corresponding time series ofelectromagnetic profiles.

92

d

e

5~

Mar 198-4

30 ~2 50TIME IN DAYS, DAY I-IS Mor 198~

2210

30 ~ 2TIME IN DAYS. DAY I-IS

spring 1985 SFISAR study plot

~ATERSATURATED LAYERS

spring 1984 Gemsluecke~ATERSATURATED LAYERS

spring 1984 SFISAR sludy plol~ATERSATURATED LAYERS

TIME IN DAYS, DAY I-IS Mar 1985

z.... 2~3--cz"o'""u~ )aa::-:xl ,.c:: -

Figure 8a-e. Damming and percolation of meltwaterin the SFISAR sludy plot and a nearby northeastfacing slope for springs 83,84,85. Theuppermost line in a-e) indicates total snowheight, the lower lines water saturated layers.

~ t.u

z _-223-

~ LG r-w1;12e:-:xl -c:: -

J::U

Q

~small

_m.dium

ITIIIIJI]]larg.

b

\, , ,\

Dllysimeter Discharg.

Dno

~0 50 60DAYS, DAY 1"15 M., 1993

~~----

TIME IN

2012

I Surfac. M.ltProduction

Dno

~SmQII

_Iargt

JJ T.mp. Profilt

I~J A

[l] B

CD C

J::U

Z C-223-cz5 t-~ Fw c~12T

t

J::U

z _-223­cz5~w -~12~:xl ­c::

~ ::­!! ~

.............~.-....-~ ............~~....ll~ .............................................,~

IIOA~spring 1983 Gem sluecke

~ATERSATURATED LAYERS

Normally surface melting starts in March. Oftenmeltwater is dammed for several weeks. In spring'83 meltwater reaching the bottom of the snowpackon the slope was delayed by about 3 weeks comparedto the study plot. The meltwater was routeddownslope along an interface or crust. In spring'84 there was almost no delay. In '85 even thoughmeltwater reached the bottom by the end of May,there remained significant damming at an interfaceabout 40cm above ground for about 50 days. InFigure 8f selected spectra from spring '86 aregiven. Again surface melting started in late March.Some damming at two interfaces can be seen in thespectrum of 2 April. Noticeable settling occurredduring April. Between the 1 May and 2 May,

spring 1983 SFISAR sludy plol~ATERSATURATED LAYERS

93

Radarprofi les Study Plot SFISRR f

Specular reflection as a function of electromagnetic distance(arbitary units)

.22V

Ie'lI.: ,m

Ierc..... ,m

r:::(',"lBr~<...~..s...e__'_19,:,:V

Ier~'..._...lea__'_19~V•18

n::~~{:",..,Ie

.3V

III.29V

18.19V

Ie.34V

III.llV

18.3~V

Ie.IIV

.3SV

j;:;;["'1':;[..

.SV

Ie1.2V

Ie1.2V

Ie1.2V

Ie•39V

Ie• S3V

• Ie.2SV

.3~V

Ie.S2V

. 'Iei:LC:

Figure 8f. Selected spectra of spring '86.

meltwater penetrated and was dammed at a lowerinterface. Some meltwater reached the bottom of thepack through channels, but the layers below thedamming interface remained mainly dry. On 6 Maywater concentrated for a short time at a lowerlevel. Until 14 May not much melting occured. Someof the water seems to be refrozen or to be routedthrough the now existing channel system to theground, leaving the main part of the pack stillalmost dry. But on the 15 May, after the onset ofsignificant meltwater production at the surface,the dammed meltwater broke through, homogeneouslywetting the rest of the snow pack.

Several conclusions can be drawn from theseobservations: Meltwater may penetrate cold snow in

preferred channels. Damming of meltwater may delayhomogeneous wetting of underlying layers for longperiods of time. This effect is even morepronounced if meltwater flows downslope alongimpermeable layers. Meltwater originating from thesurface may reach the ground before the snow coveris isothermal.

STRATIGRAPHY PROFILING

Local variability of the stratigraphysignificantly effects slab releases. Forests areknown as an effective protection against slabreleases. But today, unfortunately, air pollutionseems to severely damage protective forests. For

94

snow surface- ... - - - - - - - - . ~

typo .now depth 205m

a

.~

I,...:---.

I II I \111111' I II 111111111 111111111\111 I

SFlSAR study plotI

CJ I\~·\ I~ ' ..~ N '"I''- '3 test nne '3

> >

t =: ~ ~

T~... N

'0~

:.'

typo .no. depth 1m

ground'b

Figure 9a,b. Examples of electromagnetic profilestaken along test lines. a) SFISAR study plot,b) larch forest at timberline.

this reason we have to define minimumspecifications for protective forests to giveforesters a tool for their decisions on where andhow to start additional protective measures. Inforests, the trees influence the microclimate, thetree crowns intercept snow, and delayed snow fallsoccur from tree branches. These facts change thestratigraphy in the vicinity of a tree compared toa free-field profile. The sledge-mounted radar isused to measure the extent of these influences. InFigure 9, three examples are given. Figure 9a showsa typical free-field situation at the SFISAR studyplot. Total snow height varies significantly alongthe test line as the result of strong drifting inthe open. Figure 9b shows the stratigraphy in atypical larch stand at timberline. The influence ofthe larch crown on the stratigraphy is small. Thelarge roughness of the ground surface seems toaffect stratigraphy more effectively. A largeramount of snow is necessary in these types offorest, compared to the free-field situation(meadows), to smooth the ground allowing for a morecontineous layering. In the free-field, strong snow

drifting may smooth the surface with less snowavailable. But snow drifting is decreased byforests. The larch stand shown is typical for aforest where slabs may release during very heavystorms. Figure 9c shows a more dense spruce standat the same location. There is significantly lesssnow, interception seems to be much more effectiveand very little layering can be found at all. Thistype of forest will normally prevent slab releases

MEASUREMENT OF AVALANCHE FLOW HEIGHT

The problem of measuring flow heights and slopeperpendicular flow speed profiles in dense flow 'avalanches was the motivation for developing FMCWradar at SFISAR. Four radars are buried inavalanche tracks at our avalanche study site on theLukmanier Pass. An example of an avalanche flowingover two radars, which are buried at a slopedistance of 16m is shown in Figure 10. The picturesshow the avalanche flowing over the static part ofthe snowcover. A rock barrier above the upper radar

95

N

~~

tree crown projection•

Figure 9c.spruce

l •')

I""V

o~ (.. /"\ .

II11 I II I I I I I I I I II III I II I II I I II II I I I I I I I I I I I I I I II I I I I I I I-I I I I I 1111111111111 114 lO.6m snow surface 1

~ ... ', " ••••,..', • -, .........., •••• ~.. I -, ," ....~ ... ~.,.., °0 ,",,, """"."," .0' 0'. '. oO" ~~~/,~'"Ip'::~~~~''';/~~~''.~~..::~~"~:!,;\'t· ~:-·~t~~:.;;~ ..~: ~·";\~"'··;:,/0i.>~-I·'t-,~~!.'~~·~~~~\·/~;".:r".!4'::...::/..~':",\"-,~",~.*,~,;",,,~>~~"'7'~c

IT¥~~t.~itj~{~:11~\:~~,~j~11l;t~~i)~titJ11;"':~~~~~~fj:;;}~~i~f~:5~tIrr&:1~'•., ,.' "!:.\.. ··1\ .. ·\·,·~·~· .";, ':".J>.:"> .............,.•\''': .. v:._; .. ;;;'....~..... :., ..' ~" r·,·~ ",~.<, ...-'\.... ,~~.,~ Cr<,.,>::.~;.:~;~~:1;~jyt' , ' .,', .:> .. ,VA";' .' ·\~:)~:i~~f~~·~~"Y'~;(.->..'" ":!~;r,-"1

i,\,.i"~"':"'\'~'\'" " , , ground ji . typo snow depth O.7m C ..

Electromagnetic profile measured in aforest near timberline.

[k Hz]

6

5

6- 2,2 m

5

First Radar

Avalanche

Time

Avalanche

Stable Snow

1-

II~Ground

I I Time

Figure 10. Electromagnetic, slope-perpendicularprofile of a dense flow avalanche measuredwith two radars at a downslope distance of16m.

96

ejects some snow in the air as can be seen on thetop plot.

CONCLUSIONS

In our op~n~on, FMCW radar is a very powerfulltool for many different qualitative andquantitative investigations of the snow cover andavalanches. Because the instrument can be buried inthe ground, it is well protected from the harshenvironmental conditions. The technique is fairlysimple and reliability proved to be very high.Many operational applications can be envisioned:Monitoring of stratigraphy, loading, settling,waterpercolation, slab releases in avalancherelease zones; monitoring of avalanche flows intracks for local warning systems; monitoring ofwaterequivalence and meltwater percolation forhydrological purposes.

ACKNOWLEDGMENT

We would like to thank M.Hiller andG.Klausegger for their continuing help to maintainthe systems and R.A. Schmidt (USDA, Fort Collins,USA) for his careful review of the manuscript.

97

REFERENCES

Ellerbruch, D.A. and H.S.Boyne (1980). Snowstratigraphy and waterequivalence measuredwith active microwave system. Symp. Snow inMotion. J. of Glaciology 26:225-233.

Gubler, H. and M.Hiller (1984). The use ofmicrowave radar in snow and avalancheresearch. Cold Region Science and Technology9:109-119.

Gubler H. ,M.Hiller and R.A.Schmidt (1985). FMCWradar for snow cover investigations. InternalReport no 627 SFISAR.

Schmidt, R.A., H.Gubler and M.Hiller (1984). Swissimprovements of the FMCW radar for snowmeasurements. Proceedings of the 52nd AnnualWestern Snow Conference, April 1984, atKetchem, Idaho.