EVALUATION OF PROTECTION MEASURES AGAINST AVALANCHES IN FORESTEDTERRAIN

Thomas Feistl1, Armin Fischer1, Peter Bebi2 and Perry Bartelt2

1Bavarian Avalanche Warning Service, Munich, Germany2 WSL Institute for Snow and Avalanche Research SLF, Davos Dorf, Switzerland

ABSTRACT: Small to medium sized avalanches release in forest gaps and open forest above roads and infrastructurein the Prealps regularly each winter. Avalanche bombing, silvicultural management and technical prevention measuressuch as glide snow tripods, snow fences and galleries can prevent road closure during the winter season. Besideshistorical data and expert knowledge, avalanche dynamics models are recently increasingly employed as additionaldanger assessment tool. Modelling small to medium sized avalanches in forested terrain requires high resolutiondigital terrain models and detailed process understanding concerning forest-avalanche interaction. Removal of snowby trees, energy loss through tree breakage and higher surface friction are the main processes that lead to earlieravalanche stopping. In this study we performed avalanche dynamics simulations for four well documented casestudy areas in the Bavarian Alps where forest influences avalanche runout distance. Wet and dry snow avalancheregimes were assumed for the south, west and north facing slopes, respectively. We found differing effect of forestson velocity and lateral spread of these avalanches depending on the flow regime, the forest stand characteristics andthe underlying terrain features. Whereas technical avalanche defense structures to hinder snow gliding and supporttree re-growth are the most promising courses of action in three cases, a dam to protect the road is the most effectivemeasure for the fourth slope. These case studies demonstrate how avalanche dynamics models can support localauthorities in facilitating the planning of optimal avalanche prevention measures.

KEYWORDS: avalanche, forest, public safety, modelling.

1. INTRODUCTION

Roads and infrastructure in mountainous regions areendangered by small to medium sized avalanches eachwinter. Avalanche commissions and local authoritieshave to guarantee public safety by establishing welladapted avalanche defense strategies. Permanenttechnical structures, protection forest management,avalanche bombing and road closure are the mostcommon safety measures in Bavaria. Economical andecological constraints, however, influence the decisionmaking process. The optimal measure depends on therisk for injuries or fatalities, terrain features, avalanchecharacteristics and forest extent and composition.

A comprehensive evaluation strategy to identify theoptimal defense measure includes field studies,analysis of historic events and avalanche dynamicssimulations. Avalanche dynamics models providevaluable information on the spatial extent and runoutdistance of avalanches for different hazard scenarios. Atpresent the focus of such model calculations is on

* Corresponding author address: Thomas Feistl,Avalanche Warning Center, Bavarian EnvironmentAgency, Munich, Germany;email: thomas.feistl@lfu.bayern.de

extreme events where small scale topography,forest-avalanche interaction and snow-cover propertiesare secondary (Vollmy, 1955; Salm, 1993). Includingvegetation effects in avalanche dynamics models canimprove simulation results and thereby the evaluationof optimal avalanche defense measures, especiallyin complex, forested terrain. A careful testing onwell documented example cases is however essentialto reduce uncertainties and establish the model’sapplication range.

Recently we have seen an increased demand forforecasts concerning the runout of small to mediumsized avalanches in Bavaria. The application ofadvanced avalanche dynamics models to predict therunout of frequent events remains an on-going researchtheme. Recent models couple high resolution terrainmodels with more physics-based approaches (Barteltet al., 2011; Buser and Bartelt, 2009, 2015). Theseapproaches are able to model both dry and wetavalanche flow regimes (Vera Valero et al., 2015; Barteltet al., 2015) that include forest-avalanche interaction(Feistl et al., 2014; Teich et al., 2012). The modelsmust include entrainment to simulate avalanche growth,and important aspect of small avalanches (Dreieret al., 2014). First studies have demonstrated the

Proceedings, International Snow Science Workshop, Breckenridge, Colorado, 2016

561

application of such approaches in pre-defined areas, forexample the dangerous avalanche slopes threateningmining operations in Chile (Vera Valero et al., 2015).A prerequisite for application is both an avalanchecadastre, as well as a network of automatic weatherstations (Wever et al., 2016). Most applications havepurposely avoided vegetated regions, concentratinginstead on high-altitude applications. How to modelfrequent avalanches in forested terrain requires furtherexperience and testing, above all to provide road andforest management with engineering and silviculturaladvice for different hazard scenarious and forestdisturbances. This is the purpose of our study.

We back-calculated avalanches on four specific frequentavalanche slopes in the Bavarian Pre-Alps. Theseavalanches endanger roads which are highly frequentedand public interest increases to keep them permanentlyopen. Safety can generally be assured by a denseprotection forest if growth conditions are supportive.However, the protection forest on these slopes is underpressure through extensive damage by game animals,uprooting by dense snow glide movements, stormbreakage, forest fires and droughts. In our modelcalculations we take such differences and evolution inthe forest cover and their effects on avalanche dynamicsinto account. Additionally snow wetness and thereforeflow regimes vary depending on exposition, altitude leveland weather conditions and are considered by applyingdifferent temperature scenarios.

Optimizing avalanche protection measures is ofgreat public interest. Local authorities, avalanchecommissions and even ecosystems, especially inhighly sensitive alpine regions profit from sustainableavalanche defense concepts. This study revealsthat the integration of snow-cover properties andforest-avalanche interaction into avalanche dynamicsmodels provides valuable information on erosion anddeposition processes and finally shows how avalancheprotection measures can be optimized.

2. Avalanche model equations

To model avalanche flow we numerically solve a systemof differential equations that is conveniently written as asingle vector equation:

∂UΦ

∂t+∂Φx

∂x+∂Φy

∂y= GΦ. (1)

Flow of the avalanche core Φ is described by nine statevariables UΦ:

UΦ = (MΦ,MΦuΦ,MΦvΦ, RΦhΦ, EΦhΦ, hΦ,

MΦwΦ, NK ,Mw)T . (2)

The vector equation Eq. 1 is defined in a horizontal X-Ycoordinate system. The elevation of the mountain profileZ(X,Y ) is specified for each (X,Y ) coordinate pair. Thisinformation is used to define the local surface (x, y, z)coordinate system with the directions x and y parallel tothe geographic coordinates X and Y . The slope-parallelavalanche velocities are uΦ = (uΦ, vΦ). The avalanchemass MΦ and flow height hΦ (volume) are tracked overtime t.

The model equations include an explicit calculation of thedispersive pressure NK which is induced by mechanicalenergy fluxes associated with the hard basal boundaryand random particle movements. The mechanicalenergy of the random movements is denoted RΦ.The dispersive pressure induces slope-perpendicularz-velocities wΦ causing an expansion of the avalanchecore. The density of the avalanche core is therefore notconstant, but changes according to the basal boundaryconditions. This modelling approach allows us tosimulate both ”Fluidized” dry avalanches and dense, wetsnow avalanches.

The model tracks the total amount of meltwaterMw entrained by the core (MΣ→w), or produced bydissipative heating (MΦ→w). Tracking phase changesfacilitates the modelling of wet avalanche flows whichare governed by lubricated sliding surfaces (Vera Valeroet al., 2015).

The components (Φx, Φy) are:

Φx =

MΦuΦ

MΦu2Φ + 1

2MΦg′hΦ

MΦuΦvΦ

RΦhΦuΦ

EΦhΦuΦ

hΦuΦ

MΦwΦuΦ

NKuΦ

MwuΦ

,

Φy =

MΦvΦ

MΦuΦvΦ

MΦv2Φ + 1

2MΦg′hΦ

RΦhΦvΦ

EΦhΦvΦ

hΦvΦ

MΦwΦvΦ

NKvΦ

MwvΦ

. (3)

The flowing avalanche is driven by the gravitationalacceleration in the tangential directions G = (Gx, Gy) =(MΦgx,MΦgy) where gx and gy are the slope-parallelgravitational accelerations in the x and y directions,respectively. The frictional resistance SΦ = (SΦx, SΦy)

Proceedings, International Snow Science Workshop, Breckenridge, Colorado, 2016

562

consists of both a Coulomb friction Sµ (coefficient µ) anda velocity dependent stress Sξ (coefficient ξ),

SΦ =uΦ

‖uΦ‖[Sµ + Sξ] . (4)

These acceleration and friction terms are the principlecomponents of the right-hand side vector GΦ

GΦ =

MΣ→Φ − MΣ→Γ − MΦ→Ψ

Gx − SΦx

Gy − SΦy

PΦ + PΣ→Φ

QΦ + QΣ→Φ + QwwΦ

NK2γPΦ − 2NwΦ/hΦ

MΣ→w + MΦ→w

. (5)

The snow entrainment rate is specified by MΣ→Φ.Splashing mass at the front of the avalanche by MΣ→Γ.Mass detrained by forest interaction by

MΦ→Ψ = K/ ‖uΦ‖ , (6)

where K depends on the tree species, stand density andsurface roughness (Feistl et al., 2014).

In this avalanche model the friction SΦ is made afunction of the energy RΦ (degree of fluidization) andwater content (lubrication). The Coulomb friction termdecreases to zero Sµ → 0 for two extreme avalancheflow regimes: dry fluidized avalanches and dense wetsnow avalanches.

The model accounts for fluidization by calculating thefree mechanical free energy of the avalanche RΦ, whichis divided into the random kinetic energy RKΦ and theconfigurational energies RVΦ ,

RΦ = RKΦ +RVΦ . (7)

The configurational energy is the potential energyresulting from a volume increase of the core; that is,the expansion of the core and therefore the degree offluidization.

To model the decrease in friction from fluidizationwe make the Coulomb stress dependant on theconfigurational energy RVΦ ,

Sµ = µ(RVΦ ,Mw)N (8)

where N is the total normal force consisting of theavalanche weight, dispersive pressure and centripetalforces. Higher configurational energies indicate lowerflow densities and therefore lower Coulomb frictionvalues.

Note that Sµ is also a function of the meltwater contentMw. High meltwater contents facilitate lubricated slidingsurfaces and therefore lower Sµ values.

The velocity dependent stress Sξ is also a function of theconfigurational energy

Sξ = ρΦg‖uΦ‖2

ξ(RVΦ ). (9)

The production of free mechanical energy PΦ, is givenby an equation containing two model parameters: theproduction parameter α and the decay parameter β, seeBuser and Bartelt (2009).

PΦ = α [SΦ · uΦ]− βRKΦ hΦ. (10)

The production parameter α defines the generation ofthe total free mechanical energy from the shear workrate [SΦ · uΦ]; the parameter β defines the decrease ofthe kinetic part RKΦ by inelastic particle interactions. Theenergy flux associated with the configurational changesis denoted PVΦ and given by

PVΦ = ζPΦ. (11)

The parameter ζ therefore determines the magnitudeof the dilatation of the flow volume under a shearingaction. When ζ = 0 there is no volume expansion (nofluidization) by shearing. The free mechanical energyproduced during entrainment is denoted PΣ→Φ.

Temperature dependent effects are introduced bytracking the depth-averaged avalanche temperature TΦ

within the flow (Vera Valero et al., 2015). Thetemperature TΦ is related to the internal heat energy EΦ

by the specific heat capacity of snow cΦ

EΦ = ρΦcΦTΦ. (12)

The avalanche temperature is governed by (1) the initialtemperature of the snow T0, (2) dissipation of kineticenergy by shearing QΦ, as well as (3) thermal energyinput from entrained snow QΣ→Φ and (4) latent heateffects from phase changes Qw (meltwater production),see Vera Valero et al. (2015).

Dissipation is the part of the shear work not beingconverted into free mechanical energy in addition to theinelastic interactions between particles that is the decayof random kinetic energy, RKΦ

QΦ = (1− α) [SΦ · uΦ] + βRKΦ hΦ. (13)

The model equations are solved using the samenumerical schemes outlined in Christen et al. (2010).The model stopping criteria used sets that the simulation

Proceedings, International Snow Science Workshop, Breckenridge, Colorado, 2016

563

Tbl. 1: Model parameter values for the four example cases. The spatial resolution for all avalanches was 2 m. Weconsider curvature effects and assume a flow density of ρ = 450 kg/m3. The density of the released and entrainedsnow was also assumed to be constant ρ = 200 kg/m3. The initial Coulomb friction value µ = 0.55 and cohesion c =100 Pa are similar for all avalanches (Bartelt et al., 2015). Entrainment specifications are: one meter of snow-coveron 1500 m.a.s.l. decreasing 10 cm per 100 m; we assume velocity driven entrainment.

Name ξ0 [m/s2] µ erodability ε α β R0

Fahrenberg 1800 0.55 0.3 0.1 0.07 0.8 2.0Hagenberg 1800 0.55 0.3/0.4 0.1/0.3 0.07 0.8 2.0Weißwand 1800 0.55 0.3 0.1 0.07 0.8 2.0Antoniberg 1000 0.55 0.4 0 0.05 1.0 2.0

stops when the moving mass is only 5% of the maximummoving mass (Christen et al., 2010). The derivationof the thermal energy and vertical motion equationsare presented at Vera Valero et al. (2015); Buser andBartelt (2015). The chosen model parameter values forthe avalanche simulations on each of the four exampleslopes are denoted in Table 1.

3. EXAMPLES

3.1 Overview

We investigated four specific example cases whereavalanches endanger roads in Bavaria. In severalsituations road closures were necessary, cutting offalpine communities from surrounding regions. Publicauthorities need to assure access of public servicessuch as fire department, police and health care. Thepressure to keep access roads open increases with thetime of closure. In the following sections we presentthe specific characteristics of each avalanche track,providing safety strategies that take forest management,bombing, technical avalanche prevention measures androad closure into account (Table 2).

3.2 Fahrenberg

The Fahrenbergs‘ steep (35◦ - 45◦) southerly slopesrise above the northern shore of lake Walchensee andavalanches endanger the national highway between thecommunity of Kochel and the village of Walchensee.This road is highly frequented by tourists in winter andthe local economy strongly depends on its opening.The Herzogstand cable car provides access to thesummit thereby crossing the avalanche track whichis investigated in this study. Damage caused bygame animals, forest fire and drought stress leadto a considerable decay of the protection forest.New potential avalanche release areas subsequentlydeveloped in the last century. Steep rocky gravelimpedes tree growth and areas with long compacted

Fig. 1: The Fahrenberg: The aerial photograph showsan avalanche event on the 17th of March 2000 (a).The avalanche ran through a narrow gully and hitthe road on several meters length before it ended inlake Walchensee. The red polygon in the excerpt ofthe Bavarian cadastral avalanche register highlights thelower avalanche path (b). Green dots denote velocityreduction bumps that were built some 50 years ago.A possible protection dam (white polygon) and theaccumulated snow from all possible avalanche releaseareas is shown in picture (c). Following our simulationswe expect deposition heights up to 6 m.

Proceedings, International Snow Science Workshop, Breckenridge, Colorado, 2016

564

Tbl. 2: Characteristics of four slopes that are regularly hit by avalanches which originate from several potential releaseareas. For each slope we present the range of potential release (M0), entrainment (MΣ→Φ) and detrainment (MΦ→Ψ)volumes. Note that the release volume M0 is usually smaller than the entrained volume MΣ→Φ as the tracks arenarrow and long. Snow build-up behind trees MΦ→Ψ varies according to the length the avalanche runs throughforested terrain and the avalanche velocity. In the wet snow case on Antoniberg, where flow velocities are small, thedetrained volume is larger than the entrained volume MΦ→Ψ > MΣ→Φ. For Weißwand we compared simulations withand without forest cover. The altitude level is specified from release to runout.

Name M0 [m3] MΣ→Φ [m3] MΦ→Ψ [m3] altitude [m.a.s.l] T0 [◦] expositionFahrenberg 1057 - 2968 1507 - 10542 571 - 3222 1600 - 800 - 5 SSEHagenberg 1752 - 2332 2299 - 4251 917 - 1742 1550 - 1000 - 10 NWWeißwand 2256 1758 - 5286 0 - 958 1400 - 770 - 5 WAntoniberg 307 - 1433 122 - 355 143 - 512 800 - 630 0 SSW

grass support snow gliding, thereby hindering forestregeneration. Several release areas on Fahrenberg weresecured with technical protection measures; snow shedsprotect the road. Additionally tree re-growth is supportedby wooden tripods and fences to stop snow gliding. Twoavalanche events reached the highway in the westernpart of Fahrenberg in the last decades, increasing thepressure for further protection measures (Fig. 1b). Theavalanche in the year 2000 was well documented andfollowed a period of intense snowfall (Fig. 1a). Wesuppose a fairly dry snowpack in the upper part of theslope heating up along the track down to 800 m.a.s.l.Potential release areas were defined by analyzing photoand video material from helicopter flights. The maximumthree day snow accumulation sum for this region wasextrapolated from data measured at a meteorologicalstation close by and used for our model calculations.

3.3 Hagenberg

Avalanches that release on the northwesterly slopesof Hagenberg endanger the only access road to thevillage of Spitzingsee, in winter a highly frequentedski sports center. The pressure on keeping the roadopen is especially high as fire department, hospital andrescue service are situated in the nearby community ofSchliersee. Several avalanche paths further down alongthe road are secured with technical defense measuresand frequent bombing. Avalanche bombing is highlycontroversial here as important protection forest might bedestroyed. In the upper, not protected part of the slopeseveral avalanches released, ran through a dense forestand reached the road in the last decades (Feistl et al.,2014).

The documented events that reached the road allreleased with dry cold snow conditions. Release areasin the lower part of the avalanche track were definedaccording to aerial photographs and expert knowledge.

Further release of avalanches in the uppermost part ofthe slope cannot be ruled out. We therefore calculatedtwo possible scenarios: 1. One avalanche with tworelease areas in the lower part of the slope whichwere defined by an analysis of the aerial photographs.In this case we assumed release with a time shift of10 seconds, such that the first avalanche triggers thesecond release area. Temperatures were particularlylow for this region and we therefore suppose highentrainment rates (erodability: 0.4; epsilon: 0.3, seeTable 1). 2. One avalanche that releases above thedocumented case on a slope where isolated trees grow.Avalanche formation in this area cannot be eliminated.

3.4 Weißwand

The westerly slope of Weißwand rises above the nationalhighway between Schneizlreuth and Bad Reichenhall.Protection forest management including shooting ofgame animals, glide-snow protection measures andavalanche fences support the re-growth of forest on alarge part of this slope. Damage by game animalsmade these silvicultural measures necessary. Oneavalanche event that reached the road is documentedin the unsecured westerly part. The question arose if thedanger for further events increased after a devastatingstorm broke a large part of the trees in the upper part ofthe slope.

We assumed various potential avalanche releaseareas on the steep upper part of the Weißwand.Forest re-growth and terrain undulations reduce thesize of these areas and separate them. In allcases forest effects are important in stopping theseavalanches. Model calculations with and without forestwere performed to characterize the situation before andafter the storm.

A dense young forest developed in the wind sheltered

Proceedings, International Snow Science Workshop, Breckenridge, Colorado, 2016

565

gully, that was previously an avalanche release area(Fig. 2). Around the gully forest can until now not fullyprevent avalanche formation but storm loss starts to becompensated by young trees. Rigorous reduction of thegame animal population shows its positive effect here.

Fig. 2: The left hand side shows the Weißwandslope after it was hit by a destructive storm. Theprotection forest was destroyed in large parts of theslope. Following this event potential release areasdeveloped (red frame on the right hand side). Thegully which is now covered with dense young forest ishighlighted with a blue ellipse.

3.5 Antoniberg

The southwesterly steep slopes above the highlyfrequented national highway between Inzell andSchneizlreuth are called “Antoniberg”. Due to the lowaltitude and mostly southern exposition wet snow eventsare to be expected regularly on this slope. The longcompacted grass underneath reduces surface frictionand is supportive for glide-snow avalanche formation(Feistl et al., 2014). Additionally rain on snow eventsare typical. New release areas were defined in forestgaps that developed recently due to damage by gameanimals and droughts. We assume the avalanchesto release independently as the release areas areseparated through terrain undulations and forest cover.The snow runs down through a shallow gully until it hitsthe road.

4. RESULTS AND DISCUSSION

4.1 Fahrenberg

Model calculations revealed that a burial of the roadfrom avalanches originating from release areas in thewestern part of the Fahrenberg is possible, especiallywith a cold and deep snow-cover. The forest in thelower part of the track has an immense influence on therunout distance of these avalanches and reduces thepotential endangered section of the highway to a few

meters. Subsequent to field assessments, simulationsand consultations of the local avalanche commissionand the cable car management two possible courses ofaction are discussed by the Bavarian avalanche serviceto ascertain safety:

1. A collection dam right above the road couldcatch the expected avalanche snow (Fig. 1b).It would be easy accessible and enough spacefor deposited snow must be guaranteed. Wecalculated a maximum deposited snow volume of32,000 m3 consisting of 17,000 m3 release volumeplus 21,000 m3 eroded snow minus 6,000 m3 snowwhich is detrained by forest. The collection damwould at least need to be six meters of height.

2. Alternatively all areas where avalanches canrelease and reach the road need to be securedwith technical defense measures. A potentialrelease area of 20,000 m2 was identified wheretechnical defense measures need to be installed.Silvicultural management in the upper part of theslope could decrease the number and area ofthese release zones and reduce the number ofcritical events. Regularly avalanche bombing fromthe cable car in the uppermost section of theslope to secure a ski path decreases potentialavalanche release volumes. Forest fire, stormbreakage, damage by game animals and barkbeetle outbreaks endanger the recovery of theforest, therefore careful management is essential.This set of protection measures would relieve thelocal avalanche commission and would assuresafety for the public, keeping nature and landscapeprotection in mind.

4.2 Hagenberg

Our simulations show the potential of avalanches toreach the road. Especially cold powder avalanches thatrelease in the upper part of the slope are not stoppedby the dense forest along the track. Forest destructiondepends on flow height, velocity, snow density, treediameter and tree species (calculated with equ. 16in Feistl et al. (2015)) and is denoted in Fig. 3a.The turbulent friction is subsequently increased and theCoulomb friction decreased dependent on the K value iftree breakage is assumed. Additionally the detrainmentparameter K was decreased to 20% of its original valueas build-up of snow behind trees is minimized. Themaximum erosion height was defined according to thealtitude level and the forest cover. We assume lesserodable snow in forested terrain where interceptionhinders snowpack accumulation. Forested areas aretherefore blueish in comparison to non-forested areas

Proceedings, International Snow Science Workshop, Breckenridge, Colorado, 2016

566

(yellow/green in Fig. 3c). The amount of snow thatis detrained by trees is relatively small due to the highvelocity of the cold snow and the reduced detrainmentrates after forest destruction (Fig. 3b).

Protection measures need to consider the important roleof the forest. Light fences that allow forest re-growth arethe best option in the uppermost part of this slope.

Fig. 3: Simulation results for the upper release areaon Hagenberg. The maximum core pressure is shownin (a). Calculated tree destruction is highlighted withblack dashes. The detrained snow which is removedby trees is shown in the second picture (b). Themaximum erodable snow-cover on the Hagenberg slopeis highlighted in (c). It depends on the altitude (1m on1500 m.a.s.l decreasing by 10 cm per 100 m) and on theforest cover (Maximum erosion height is a function of theforest parameter K.).

4.3 Weißwand

Our model calculations underline the protective effectof the forest. Forest destruction by the storm wascompensated with high roughness of the dead wood andstumps which were kept in the release zone. Recentlyre-growth of young trees partly fulfill the protectivefunction in these areas, especially in the distinctive gully.Rapid re-growth of the young forest in the lower part ofthe slope is expected to ensure safety for the road in thefuture. Temporal technical defense measures, however,need to be taken into account.

4.4 Antoniberg

Two possible events where the road gets hit by awet snow avalanche were revealed by the modelcalculations (Fig. 4). These are the avalanches

with the largest release volumes. We are convincedthat a healthy protection forest would hinder avalancheformation, reduce release area size and thereforeprevent avalanches to reach the road. Conditions forforest regeneration are generally good; however, gameanimals and gliding snow suppress young tree growthand undermine forest management interventions. Wesuggest therefore a considerable reduction of the gameanimal population and fences to protect the youngplantations. Glide-snow is an issue and should bereduced for example with wooden tripods. Temporaltechnical defense measures need to be considered ifroad opening needs to be assured.

Fig. 4: We present the simulations of two avalanchesthat release on Antoniberg and reach the road. Wealways assume wet snow conditions to take expositionand altitude level of this slope into account. Severalother potential release areas are too small to reach thehighway.

5. CONCLUSIONS

To evaluate avalanche protection measures againstsmall to medium sized avalanches that run throughforested terrain, avalanche flow dynamics, forestcomposition and small-scale terrain features must betaken into account. Avalanche dynamics simulations area helpful tool to evaluate different hazard scenarios andquantify the effectiveness of different mitigation methods.Model applications highlight the need to simulate:

• snow interception by trees which lowersentrainable snow heights in forest terrain,

• the amount of snow that is stopped by trees. Thisstrongly depends on forest stand characteristics,surface friction and avalanche flow regime.

• the change in avalanche behaviour when trees arebroken and entrained in the flow

Proceedings, International Snow Science Workshop, Breckenridge, Colorado, 2016

567

• the avalanche flow regime which changes fromdry to wet snow, from fast (fluidized) to slow (butlubricated). Thus, avalanche phase transitionshave an influence on forest detrainment rates andtree breakage.

These processes have been parameterized and includedin our model calculations. Clearly, model applicationsrequire the specification of a new set of snow-coverboundary conditions, including snow-cover heightsand temperature distribution in forested (low-elevation)terrain. Our case studies show that engineers needto evaluate various scenarios and only knowledgeon individual processes can guarantee an optimalavalanche defense strategy for a specific slope.Comparison of model calculations to documentedevents, both dry and wet, therefore remains a priority inavalanche practice. Each comparison would help identifyhow our understanding of avalanche flow in forests canbe improved in order to evaluate mitigation methodsand the role of ecological and manmade disturbancesin forest-avalanche practice.

ACKNOWLEDGEMENTS

The authors thank Marc Christen and theRAMMS development team for many helpful newimplementations, such as the ability to specify snowinterception in forest terrain and modelling treebreakage. This work was funded by the BavarianEnvironment Agency.

REFERENCES

Bartelt, P., O. Buser, Y. Buhler, L. Dreier, and M. Christen, 2015:Numerical simulation of snow avalanches: Modelling dilatativeprocesses with cohesion in rapid granular shear flows. AnalyticalMethods in Petroleum Upstream Applications, 327.

Bartelt, P., L. Meier, and O. Buser, 2011: Snow avalanche flow-regime

transitions induced by mass and random kinetic energy fluxes.Annals of Glaciology, 52, 159–164.

Buser, O., and P. Bartelt, 2009: Production and decay of random kineticenergy in granular snow avalanches. Journal of Glaciology, 55,3–12.

Buser, O., and P. Bartelt, 2015: An energy-based method to calculatestreamwise density variations in snow avalanches. Journal ofGlaciology, 61, 563–575.

Christen, M., J. Kowalski, and P. Bartelt, 2010: RAMMS: Numericalsimulation of dense snow avalanches in three-dimensional terrain.Cold Regions Science and Technology, 63, 1–14.

Dreier, L., Y. Buhler, W. Steinkogler, T. Feistl, and P. Bartelt, 2014:Modelling small and frequent avalanches. In: Proceedings of theInternational Snow Science Workshop, Banff, Canada.

Feistl, T., P. Bebi, M. Christen, S. Margreth, L. Diefenbach, andP. Bartelt, 2015: Forest damage and snow avalanche flow regime.Natural Hazards and Earth System Sciences Discussions, 3,535–574.

Feistl, T., P. Bebi, L. Dreier, M. Hanewinkel, and P. Bartelt,2014: Quantification of basal friction for technical and silviculturalglide-snow avalanche mitigation measures. Natural Hazards andEarth System Sciences, 14, 2921–2931.

Feistl, T., P. Bebi, M. Teich, Y. Buhler, M. Christen, K. Thuro, andP. Bartelt, 2014: Observations and modeling of the braking effect offorests on small and medium avalanches. Journal of Glaciology, 60,124–138.

Salm, B., 1993: Flow, flow transition and runout distances of flowingavalanches. Annals of Glaciology, 18, 221–221.

Teich, M., P. Bartelt, A. Gret-Regamey, and P. Bebi, 2012: Snowavalanches in forested terrain: influence of forest parameters,topography and avalanche characteristics on runout distance.Arctic, Antarctic, and Alpine Research, 44 (4), 509–519.

Vera Valero, C., K. Wikstroem Jones, Y. Buhler, and P. Bartelt, 2015:Release temperature, snow-cover entrainment and the thermal flowregime of snow avalanches. Journal of Glaciology, 61, 173–184.

Vollmy, A., 1955: Uber die Zerstorungskraft von Lawinen. SchweizerBauzeitung, 73(12/15/17/19), 159–162, 212–217, 246–249.

Wever, N., C. Vera Valero, and C. Fierz, 2016: Assessing wetsnow avalanche activity using detailed physics based snowpacksimulations. Geophysical Research Letters.

Proceedings, International Snow Science Workshop, Breckenridge, Colorado, 2016

568