AN OVERVIEW OF SNOW AND AVALANCHE RESEARCH IN INDIAN HIMALAYA

IS. S. Sharma*Snow & Avalanche Study Establishment, Manali (H.P.) India

ABSTRACT: Snow and avalanche research in India took its roots during early sixties. ThoughIndian Himalaya have a very large area which experiences seasonal snow cover and permanentsnow-bed, SASE has oriented its research to study problems related to specific areas of relevance.Indian Himalaya, experience such a wide diversity in climatic and precipitation patterns that the snowproperties and related avalanche activity assume a wide variation. This has necessitated orientationof its research programme, which is in slight variance with what is being followed in Europe and NorthAmerica.

In the field of Snow Physics and Mechanics, we are engaged in studying properties of snow inthe range of -10°C to -3°C and relating such studies to microstructure. In the field of avalancheforecasting, we are engaged in developing a Nearest-Neighbour model and an expert system forsome specific areas while using a process oriented approach for the general area. In the field ofavalanche dynamics we are concentrating on development of dynamic models for wet and semi-wetsnow avalanches taking cohesion as an important friction parameter. In the field of avalanche controlstructures, we are engaged in developing design criteria for retaining barriers encountering high­density semi-wet snow with large depths and rapid settlement. For assessment of snow coverproperties of a large area through satellite imageries, we are engaged in developing suitablealgorithms for change detection, pattern recognition and snow characteristic classification.

The paper describes the snow and avalanche problem with respect to Western Himalaya anddiscusses some specific areas where snow and avalanche related research is being done at SASE.

KEYWORDS: Avalanche climate, Himalayan snow, Indian Avalanche, Indian Himalaya.

1. INTRODUCTION

The avalanche affected area in IndianHimalaya lies in Western Himalaya, CentralHimalaya and North Eastern Himalaya. Theabove area is spread over a distance of 2500­km covering 26° N to 37° N latitudes and 72° Eto 96° E longitudes. Of these areas, the areafalling under Western Himalaya covering thestates of Jammu & Kashmir (J&K) and HimachalPradesh extending from 31° N to 36° N latitudes,73° E to 80° E longitudes and altitudes 2000 m to7000 m is most populated and most frequentedarea which has varying nature of terrain. Thisarea also experiences heavy snowfall andavalanche activity. The complex nature ofterrain, its proximity to sub-tropics, high altitudeand heavy snow precipitation due to WesterlyDisturbances warrant special snow andavalanche research programmes which arespecially suited for Indian conditions.

*Corresponding author address: S S Sharma,Snow & Avalanche Study Estt (RDC),3167/24D, CHANDIGARH -160 023 (India)Tel: 0172-713167 & 709095; fax: 710216;email: sss@sasehq.ernetin

558

2. CLASSIFICATION OF SNOW CLIMATE INGENERAL

For the purpose of classifying .theavalanche actiVity with respect to weatherpattern, McClung arid Schaerer (1993) haveclassified the avalanche areas into twocategories, as Maritime and Continental. As perthem, relatively heavy snowfall and mildtemperatures characterize the maritime snowclimate. Avalanche formation takes place duringor immediately following storms, with failuresoccurring in the new snow near the surface.Prediction of such types of avalanches based onsnow observations is fairly accurate.

On the other hand, continental snoWclimate is characterized. by relatively lesssnowfall, cold temperatures and the locationsare considerably inland from the coastal areas.Snow cover is relatively shallow and oftenunstable due to persistence of structuralweaknesses. The avalanche frequency isgenerally low and the low temperatures.generally allow structural weaknesses to persistfor longer duration. Prediction of such

avalanches is rather complex and requirescontinuous monitoring and evaluation.

3. CLASSIFICATION OF SNOW CLIMATEFOR WESTERN HIMALAYA

Table 1: Terrain and Meteorology of IndianHimalaya

Lower Middle UpperFactors Himalayan Himalayan Himalayan

Zone Zone ZoneTerrainAltitude 3200- 3500- 5000-

4100m 5300m 5600m(76%) (100%) (100%)

Slope 30-38 32-40 28-32(64%) (75%) (67%)

Ground Tall grassy Scree and Rocky,cover boulders scree and

glacialMeteorologySnowfall in a 20-80 em 20-80 em 10-20 emstorm (56%) (81%) (51%)Average totalyearly 15-18 m 12-15 m 7-8 msnowfallTemperature("e)Highest max 20.2 14.5 9.0Mean max 6.8 0.96 -8.1Mean min -1.6 -11.3 -27.7Lowest min -12 -25.4 -41

From climate and avalanche actiVitypoint of view, Sharma and Ganju (1999) haveclassified the Western Himalaya in three zonesas, Lower Himalayan Zone or Subtropical Zone,Middle Himalayan Zone or Mid Latitudinal Zoneand Upper Himalayan Zone or High latitudinalZone as shown in Figure 1. A brief description ofeach of them is given below:

3.1 Lower Himalayan Zone or SubtropicalZone

This zone could be classified as thezone of warm temperature, .high precipitationand short winter periods of three months. Theprecipitation is generally concentrated betweenDecember-March with the periods before andafter experiencing wet snow precipitation orrains. The snow cover very soon changes intoisothermal snow pack. .

7!5 78

Ali,(m)119224803'0503800158028003691308026503JZ333142%02972:31803100244024144040275021453200476042502%03250381)0

l,egendS~tIoJlN:lme1. Sin!~. Sm~

3. 8m34. Stn4$. SUl$6, Stn67. Stn78.8tn89. Silt')

10$inl0U-SinU12;SmHHSinH14.S11114IS.StlllSl6.SIJiJ617.8tn.1118.S'1I1819.5101920.SUl2021.5tn2122'stn'UAlStn2~

24£tn24ZS$tn25265m26

33

·32

-)1""'-':-

/J \ 't

( ~j....--._"':\ r-.,...-J

) '.' ..-L..,-----.-;.....-- --"'--.... ~)

1,--.---",...'INDLA

...._ i

-" !

Ml\NDr~J"),

~,...~.

~ .''...,..C;3ANDlGARH

: RepresentatiYe ObSCf'iatOI:y

Fig.1: Snow-avalanche climatic zones and the network of observatories in Western Himalayanregion.

559

The avalanche activity is quite high, withmost of the avalanches triggering duringsnowfall as direct action avalanches due toexcessive overburden, or within 24 hours after amajor snowfall on a clear sunny day. The peakwinter avalanches are generally moist slabavalanches and late winter avalanches are meltavalanches (thaw avalanches) containing snow,mud and stones.

The mountainous areas falling in thiscategory are the Pir Panjal range in Jammu andKashmir and lower altitudes on the windwardside of the same range in Himachal Pradesh.These areas have tree line up to 3000 m and areheavily populated because of prevalence ofpleasant climatic conditions.

3.2 The Middle Himalayan Zone or MidLatitudinal Zone

The middle Himalayan zone ischaracterised by the high mountain peaks andnumerous glaciers. The terrain is rugged and isgenerally devoid of vegetation except in a· fewpreferred areas where slopes up to 3000 maresparsely forested. The zone has ma~imum

avalanche slopes in 3500 - 5300 m range.The areas falling in this zone are

windward side of the Great Himalayan range inJammu and Kashmir and upper reaches of PirPanjal range in Himachal Pradesh. This zone issparsely populated by virtue of being rugged,cold and mostly glaciated. This range receives .good amount of total snowfall during winter, 80%of which is through moderate snow spells of 20­80 cm. Entire middle Himalayan range receivesdry snow between mid December and endJanuary. The rise in temperature from midFebruary onwards generally moistens freshsnowfall and after March the fresh snowfall isoften accompanied with light rain or wet snowprecipitation. Severe wind activity redistributessnow from avalanche slopes very frequently inthis zone.

Severe avalanche activity is reported inthis range throughout the winter. The massiveslab avalanches from drift loaded slopes arealso observed. Thaw avalanche activity is alsoobserved from a few slopes in the months ofApril and May.

3.3 Upper Himalayan or High Latitudinal Zone

The average height of this zone is about5000 m and it houses some of the longest

560

glaciers of the world. The areas falling in thiscategory are a few slopes in the leeward side ofthe Great Himalayan range (Jammu andKashmir and Himachal Pradesh), Zanskar rangeand Karakoram range.

This part of the Himalaya is very thinlypopulated. The climatic conditions at someplaces in winter are closer to polar conditions.Snowfall in this zone is generally scanty but it isextended almost throughout the year. Thesnowfall is mostly dry and bonds poorly with theglaciated surface or with old snow. The· totalprecipitation as well as the precipitation intensityremains low in this region. However, whatever

. little precipitation that takes place remains forlonger duration till the melt season starts in May.Since snow on slopes remains mostly looselybonded, the redistribution due to wind activitytakes place very frequently.

Avalanches occur from steep slopes inthis region, however, their frequency is not veryhigh. Since the ground conditions are notconducive to anchor the snow pack, avalanchesfrom glaciated and steep rocky surfaces startwith as little as 30-40 cm of fresh snow. Incertain areas delayed action avalanches arealso observed.

4. SNOW RESEARCH

Keeping the above conditions in mindsome special research programmes in hand aredescribed below:

5. SNOW PHYSICS AND MECHANICS

The research work has been just startedon modelling wet snow grains and bond growth.and bond dissolution. Bond distribution of wet

. snow is also being monitored with varying cyclictemperatures in cold laboratory. Process of bondformation has been monitored with the help ofthin sectioning under polarized light (SatYawaliand Sinha, 2000). In field, a high-resolutionpenetrometer is expected to give some moreinformation with regard to the bond formationand bond dissolution. Martin Schneebeli at SLF(personal communication) finds some excitingresults using high-resolution penetrometer in thefield during melt period.

Constant load creep experiments at hightemperatures exhibit that creep-rate is verysensitive to temperature level in the range of-10°C to -2°C as shown in Figure 2. Iftemperature is varied (from -5 °c to 0 0c) in a

6. SNOW COVER SIMULATION MODEL

\.-

-3 oC

100 500

at different

40 1~ 35 1~ 30~

0, 25c[I! 20

U5~ 15

.~ 10.Q)

.1- 5

0+-~~~-~~~~~~~~~~--4

2 5 10 40 55

Strain-rate (10'5 S·1)

Figure 4: Critical strain ratetemperatures.

cycle during the experiment, axial viscositydecreases with increase in temperature. Alsopercentage increase and decrease in axialviscosity decreases with successive cycles(Refer. Figure 3). Critical strain rate (ductile­brittle transition) at _3°C was determined to be10% of that at -10°C for a snow density of 400­kg.m-3 (Refer Figure 4). Critical strain-rate mayprove to be the most important failure criterion topredict wet snow avalanches. A programme hasbeen initiated to conduct extensive experimentson high temperature snow and to corroboratethe results with the field data particularlyacquired during the melting season. Thedependence of mechanical behaviour onmicrostructure will be given more emphasis inthis work.

E~I~=;~~..s-10 .., \ -10oCc

~ ' 15 1' \.E Density = 450 Kg m '" -3oC I.E I Load = 0.1327 KN "-------. '~ -20 -1 Sample diameter = 65 mm :

. I-25 +-!~,.-,-~,.-,-~,.-,-~,.-,-~.,......,...~~~-,--,-~..,..-j

First attempt to model physical processes ofthe snow cover were initiated by Ganju et. at.(1994). This model was mainly based on thedensification laws to predict the snow height atvarious locations. In 1995 a new scheme hasbeen started to simulate the snow cover of thethree zones. The model is a one-dimensionalmodel. It calculates heat conduction, settlement,phase change and water percolation in the innerlayers. Metamorphic processes are recentlybeing taken care of. The basic flow proceduresof the snow pack are displayed in Figure 5.

Figure 2: Creeping behaviour of snow atdifferent temperatures under compression tests.

3.18 737.37 2502.24 10810.21 183844.96

Time (Secs)

SASEstation

II AWSdata II Armysta60n II I

Local Meteorologicalconditions

Figure 3: Axial viscosity with cyclic temperature.

4.5 ,---------------~_3"..,Density=399 kg m

!if 4m All. 3.5 / '"~ 3 Stress=40 KPa j',' ',-

< I ' ...

~~1 . . ,~ 2 /~"------:J&// "',~1 5 ////-..-----.....~. Density=350 kg m"

:~ '1 j /{/> 0.5

Ol---~--~-~--~--~--

-,,,~"i, '___________ I

Day... ;.. :I

Snow pack structure(metamorphism)

Temperature, density. and liquid water content

oo ~ 0 ~

Temperature (0C)

-5

Figure 5: Basic components of the one­dimensional snow pack model.

561

6.1 Test Results

.The model was tested on the datacollected at two well-instrumented observatoriesof SASE representing diverse topographic andclimatic conditions (Singh et. al. 1999) (ReferFigure 6). The first station Dhundi is located in avalley at an altitude of 3000 m in Pir Panjalrange. The station experiences moderately hightemperatures, intense snowfall and moderatewinds.

The second station Patsio is located inGreater Himalayan range between middleHimalayan and upper Himalayan zone, at analtitude of about 3800 m above sea level, on aflat topography· where three valleys meettogether. The station experiences high wind,relatively less snowfall arid low temperatures.

Figure 6: Snowcover evolution at Dhundi andPatsio.

The salient points observed during1998-99 winter are:(a) Snowpack was simulated very well for themiddle Himalayan zone.

562

(b) Snowpack in the lower Himalayan zonewas underestimated during melting period.(c) Formation of TG snow retards densification.Model could not capture this effect very well.(d) Overall, extreme low and warm conditionsare the two major factors, which are to befurther studied in detail.

7. AVALANCHE FORECASTING

For covering the large area we follow themethod of process oriented approach toavalanche forecasting. However, for specificareas where we have a dense network ofobservatories and where winter movements aretoo frequent, we are in the process of developingsome numerical techniques for achieving higheraccuracy for avalanche forecasting.

7. 1 Expert SYstem

SASE uses an expert system for theChowkibal-Tangdhar road-axis, which wasdeveloped with the help of an expert shell, C­Language Integrated Production System(CLIPS). The rules are formed with28 snow-metand snow profile variables derived from sevenwinters data of an observatory situated on theaxis (Naresh et. al. 1999). The classification ofwinters directly refers to the variation of thesnow and meteorological conditions from earlywinter to late winter. This has led to thedevelopment of rules based on the variables foreach winter month starting from December toApril. The site-wise rules, which deal with therelease of avalanches, can be termed as a sub­set of rules that control the avalanche days.Therefore, these rules are activated only when aday is declared as an avalanche day by the mainset of rules. .

The rule base consists of 358 rules. outof these 173 rules are decision rules. The modelgenerates avalanche goals for separateavalanche sites with separate possibility valuesbased on a given day's fact set. In this way, 24­hour forecasts are generated by the expertsystem.

Table 2. Results of the 24-hour forecast for theperiod 16 January - 6 April 2000 (87 days).

# . Category PredictedPredicted Mis-correct! Classified

1 Avalanche Days 38 11 28

2Non-Avalanche 49 49 0

Days

3 Percentage 100 69 31

From Table 2, it is seen that the non­avalanche days predicted by the model wereobserved as true non-avalanche days. The mis­classification in case of the number of avalanchedays is mainly due to non-reporting ofoccurrence from the areas, which are not underdirect observation.

7.2 Nearest Neighbour Method

We are also utilising ·the numericaltechnique of Nearest Neighbour Method foravalanche forecast. This technique involves thecomparison of current day's snow-met conditionwith that of past days' to select a set of dayshaving most similar characteristics. Forecasterthen analyses the events associated with eachselected day and draws an inference about thelikelihood of event that may take place on thecurrent day under the prevailing snow-metcondition. This technique is extensively beingused to predict avalanches of a highly avalancheprone glaciated region in Upper Himalayan zonein Karakoram Range. The data collected at 0830h each day is used for calculations. Havinganalysed the events associated with eachselected day, site-wise prediction is then madein terms of probability. A record of predicted andobserved avalanche activities for 1998-99 winterin this region is. presented in Table 3. Theinformation gathered through this method assiststhe forecaster in preparing the final forecast.

Table 3: Number of days predicted andobserved avalanche occurrences by NearestNeighbour Method - 181 days

Number Predicted Probability (%)

ofdays 0 10 20· 30 40 50 60 70

Predicted 97 24 22 18 10 04 05 01

Actual 142 03 12 09 06 03 05 01

8. AVALANCHE DYNAMICS

The avalanche flow parameters aredirectly related to its initial condition like fracturearea and fracture depth. According to SwissgUidelines, the fracture depth of a formationzone depends on three days maximum increaseof snow cover depth. However, we haveobserved fracture depth of the order of 1.0 - 2.5m in the Lower Himalayan Zone and 0.5 - 1.5 min the Middle Himalayan Zone with a high

563

frequency of the order of 1 - 3 avalancheoccurrences in a winter.

SASE has developed its model basedon Saint Venant type of differential equations forcontinuity, momentum and energy, where theeffects of entrainment, detrainment, bends,density variation have been considered (Kumaret. a/. 1998). The study of the effect of cohesionin snow mass is also taken into account.Entrainment of snow mass locally increases the'resistance to the avalanche flow due toexchange of momentum, and if the avalanchesize is small it can stop the snow mass byreducing the avalanche velocity. The detailedstudy for the effect of cohesion has been carriedout. The important conclusions are: (a) Cohesionclearly introduces a size effect into avalanchedynamics calculation (Bartlet et. al. 1999). Thesmaller avalanche will run shorter distances andstop on the steep slopes and large avalanchewill run longer distances even on flat slopes. (b)In comparison to the Voellmy fluid model, thecohesive model is more sensitive to change inchannel flow width. This is due to the fact thatflow height increases in narrow channels and thecohesive resistance is inversely proportional toflow height, which produces less resistance tothe flow. This fact helps to explain the "train-like"motion of a wet snow avalanche. (c) Cohesioncan rectify two discrepancies of the Swissguidelines Le. effect of avalanche size and wetsnow avalanche.

9. CONTROL STRUCTURES

9.1 Formation Zone Control

About 2000 m supporting structures inthe form of Snow rakes, Snow bridges and snownets have been installed in one of the avalanchepaths located in Pir Panjal range at an altitude of3000 m in the Lower Himalayan Zone. For thelast 10 years, the studies on the effect ofsupporting structures are being carried out. Inthis location throughout the ridge, cornice of theheight of 5.0 - 6.0 m is formed, breaking ofwhich has caused some damages to supportingstructures as well as excessive accumulation ofsnow over the structures. In the last three years,the Jet roofs have been installed to prevent thecornice formation and after its installation noformation of avalanches has been observedfrom the area of formation zone controlstructures. The design force for snow packheight 3.0 - 4.0 m varies between 60 kN m-1 to

100 kNm-1 as per measurement of snowpressure on supporting structures which iscomparatively higher than the values obtainedfrom the Swiss guidelines.

9.2 Middle Zone and Runout Zone control

SASE has designed and constructedearthen diversion dams at two locations. One on0-11 avalanche site to protect National Highway(Jammu - Srinagar) and other in Badrinath areato protect the holy Badrinath Temple and avillage nearby. The dimensions of diversion damon 0-11 avalanche site are height 7.6m andlength.75.0m, and in Badrinath area, thedimensions are height 6.5m and length 204m.The diversion dam in Badrinath area constructedon a ground slope of 14° was designed towithstand the avalanche striking the dam sectionat a velocity of 11.6 ms-1 at an angle of 17°. Theearthen diversionary dam was designed for aheight of 6.5 m with upstream slope of 1:1.3 anddownstream slope of 1:1.75. These structuresare working satisfactorily for over 15 years. Dueto frequent occurrence of wet snow avalanche inthese areas the transported rocks and bouldershave reduced the effective height by one meterwhich needs maintenance work.

Recently, design of a snow gallery onJammu - Srinagar highway of length 410m andwidth 10m has been carried out. The slope ofavalanche path at the road location is 14° to 17°,where design avalanche velocity and flow depthare 25.6 ms-1 and 6.72m respectively. Thegallery design is based on normal force of 5.65tm-2 and shear force of 2.35 tm-2

.

10. SATELLITE IMAGERY FOR AVALANCHESTUDIES

10.1 Snow Cover Area Estimation

Accumulation and ablation pattern of theseasonal pattern of snow cover of an area inJ&K was studied using IRS WiFS images for aperiod' between October 1996 and May 2000.Five elevation zones at an interval of 600 m from2600-m to 5600-m were created and seasonalsnow cover in individual altitude zones wasdelineated. Algorithms have been developed toquantify the snow cover accumulation at variouselevation zones during various phases of awinter. Studies for further classifying the old andnew snow are in hand.

564

10.2 Snow Cover Quality Analysis

Multi-date IRS 1C/1D PAN and L1SS IIIdata for a glaciated area have been analysed foridentification and determination of various snowparameters, such as areal extent of snow cover,differentiation between fresh snow, old snow andice. Using multi-date satellite data and alsoRadarsat (microwave SAR) data, it is possible tocarry out change detection studies of variousidentifiable features/phenomenon.

10.3 Virtual Reality for Snow Cover Studies

Use of Virtual Reality (VR) technique for3D visualization as well as for creation of walk­throughlfly-through models of avalanche proneslopes with details of terrain features includingforest cover, shrubs, snow cover can improvethe capability of assessing the behaviour ofaslope. This work is being carried out for an areain J&K using multi-resolution/date satellite andancillary data.

11 CONCLUSION

The above text covers some of the specialareas where snow research for solving specificproblems related to Indian Himalaya are in handfor futuristic use. However, for solving our on­going problems,we are utilising the work alreadycarried out by various research institutions of theworld, while updating the above, time to timethrough the inputs received through our work. Itis felt that the work in hand in some areas maybe of common interest to the community of thesnow scientists of various countries.

12 REFERENCES

Ganju, A, Agrawal, K.C. and Rao, D.L.S., 1994:Snowcover Model. Proc. Snowsymp 94,Manali, 26-28 Sep 1994, 394-413.

Kumar, A, Sharma, S.S., Mathur, P., 1998:Numerical Modelling of Avalanche Flow. InHestnes, E. ed. 25 years of Snow andAvalanche Research. Proc. VossConference, 12-16 May 1998, NorwegianGeotechnical Institute Publication No. 303,Oslo Norway, 160-164.

Bartlet, P., Kumar, A, 1999: A Dense SnowAvalanche Model with Cohesive Plug and

Fluidised Layer Regimes, SLF internal report(unpublished).

McClung, D. Schaerer, P., 1993: The AvalancheHandbook: Published by the Mountaineers1001 SW Klickitat Way, Seattle andWashington 98134. P 17, 18.

Naresh P, Pant. L.M., 1999: Knowledge-basedsystem for forecasting avalanches ofChowkibal-Tangdhar axis (J&K). Def Sc.Journ India. Oct 1999. Vol. 49 No 5. 381-.391.

Satyawali, P.K., and Sinha, NK, 2000:Microstructure: A possible tool to stUdymetamorphism and material properties ofsnow. (Being submitted to CRST, Elsevier,Netherlands.)

565

SASE Annual Reports (1971 to till date).

Sharma, S.S., Ganju, A, 1999: Complextiies ofAvalanche Forecasting in Western Himalaya- An Overview. Journ. of CRST Elsevier,Netherlands

Singh, A, Ganju, A, Satyawali, P.K., Sethi,D.N., 1999: Snowcover simulation model: Acomparison with snowpacks of Pir Panjaland Great Himalayan ranges. Proceedingsof the National Snow Science Workshop(NSSW99) Manali, India (under print).