Avalanche accidents in back cuntry terrain fthe Swiss alps: New investigatins f a 3 year database Stephan Harvey *, Chatrigna Signrell Swiss Federal Institute fr Snw and Avalanche Research (SLF), Davs Manuel Genswein Rescue and Survival Abstract: Mre than 1 avalanche incidents invlving skiers, snwbarders and muntaineers in the back cuntry terrain f the Swiss Alps have been filed in an electrnic database. New results cncerning accidents n ski-turs and ut fbund-skiing (ff-piste) can be presented. 95 % ff all avalanche accidents were human triggered. In 6 % f the cases the first persn in the slpe triggered the avalanche. In 3 % fthe accidents ski tracks frm the same day fthe avalanche were present in the slpe, in 1 % lder ski tracks were fund in the slpe. Althugh the same amunt faccidents n ski-turs ccurred n descent as n ascent, 5 % mre peple gt caught while ascending. The typical victim is male, between 17 and 3 years ld. Fatal avalanches are larger than thse where nbdy gt killed. Analysis f cmplete burials shws that peple are mre affected n multiple burials than expected. On 24 days in the last 3 years 4 r mre avalanche accidents happened n a single day. Many accident -days ccurred after new snw, strng winds with weak snw pack layers. In 2 % f the days there was neither a significant amunt f new snw, nr strng winds, but an increase f temperature. On these days the mean predicted avalanche danger was the lwest. Keywrds: avalanche accident, avalanche incident, avalanche accident statistics, avalanche frecast, avalanche bulletin, avalanche danger degree. 1. Intrductin Over the last years an avalanche data base was develped at the Swiss Federal Institute fr Snw and Avalanche Research (SFISAR). This unique data base cntains all knwn naturally triggered avalanches causing damage ver 1 years and all human triggered avalanche incidents in back cuntry terrain f the last 3 years. It cntains ttally rund 11 ' datasets. The statistics presented here draw n the data base f all recreatinal avalanche accidents frm 197 t 1999. This kind f avalanches includes all avalanche accidents where peple n tur r ff-piste (e.g. skiers, snwbarders, muntaineers, etc.) gt caught. Ttally 2649 peple were invlved (677 died, 1986 survived) in 157 separate incidents. 95 % f them were human triggered,s % naturally. The avalanche accidents were analysed in a first part with Crrespnding authr address: Stephan Harvey, Swiss Federal Institute fr Snw and Avalanche Research, Fliielastrasse 11, CH-726 Davs Drf, Switzerland; tel: +41 81 417 1 29; fax: +41 81 417 1 1; email: harvey@slf.ch. a fcus n new evaluatins. In a secnd part a cmparisn f the recreatinal accidents with snw and weather parameters f bservatin statins frm the SFISAR is presented. The results base n studies fsignrell (21). 2. New results f recreatinal avalanche accidents 2.1 SpatiaL and tempral distributin An analysis f the areas in Switzerland where recreatinal accidents ccurred, shws that in the regin fdavs much mre happened than in all the ther larger ski resrts f the cuntry (Signrell, 21). Whereas the number f fatal incidents in Davs des nt shw this tendency. This can be explained by the fact that the SFISAR gets t knw f nearly all avalanche incidents in the regin, whereas frm ther areas the infrmatin f nnfatal avalanches is nt always transferred t the institute. Therefre we assume that there is a large. estimated number f unknwn cases were peple gt caught, but nbdy was badly injured. With the cmparisn f fatal accidents between Davs and ther ski resrts a estimated number f unknwn
Internatinal Snw Science Wrkshp (22: Pentictn, B.C.) nn-fatal cases f abut 65 % was estimated This percentage turned ut t be a little higher than the estimatin by Tschirky et al. (2) (smewhat belw 5 %). Since 197 93 % f all recreatinal avalanche accidents ccur during the winter mnths Nvember t April (Figure 1), 1 % mre than in Canada (Jamiesn and Geldsetzer, 1996). 5 % f all accidents happen n a weekend day (Saturday r Sunday). During the week there are n significant differences in the days (Figure 2). 35 (J) Q) (J) 25 OJ u a 2 Q; 15. E :::J z 1 5 ::::r 4 t,." :: ', -' k -. n g :f % ] i?'?' -'= c..a 2: LL Figure 1: Number f recreatinal accidents fr each mnth frm 197 t 1999. 4 VI., 36.. VI 26-2., 16. 1 E ::> 6 z >-- >--,.., >-- >-- c, Ii l ] i E E 8 Z ::; =.,., il c t=- = >-- ""' c, (J) Figur 2: Number f recreatinal accidents fr each weekday frm 197 t 1999. 2.2 Ttal burials Frm 2649 caught peple in recreatinal avalanche accidents 698 were cmpletely buried and culd nt be fund due t visible parts. T further develpment f avalanche transceivers and fr rescue teams, statistics with this number f burials are f interest. Burial depth f all cmpletely buried peple are deeper than in investigatins f Tschiky (2), because the presented results in this paper relay n a database cntaining peple wh were nt fund due t visible parts. The median burial depth f all 698 cmpletely buried peple is 1 em. The depth fr survival peple is 7 em, the ne fr dead 12 em. Cmpletely buried peple (698 cases) All (tur: 471; Dead (tur: Survived (tur: ff-piste: 227 342; ff-piste: 128; ff-piste: cases) 144 cases) 83 cases 1 em 12 em 7 em Table 1: Median burial depth f cmpletely buried peple fund withut visible parts. Burial depth fr turing and ff-piste terrain are the same. 27 % f all avalanche accidents, where peple were nt fund due t visible parts, include multiple burials. Fr accidents n turs 35 % were multiple burials, fr ff-piste 16 %. Therefre in tw f tree avalanche accidents n turs nly ne persn is ttally buried. In casualties pint fview, 61 % fall buried peple n turs are invlved in multiple burials. Apprximately 1;4 are buried with 4 r mre peple. In ff-piste terrain arund 3 % are invlved in multiple burials and 4 % are buried with 4 r mre peple (Table 2 and 3). 45
Turs Off-piste Ttal Rescue and Survival Number f Number f Number f Number f Number f Number f Number f burials Accidents affected peple Accidents affected peple Accidents affected peple 1 183 183 156 156 339 339 2 49 98 23 46 72 144 3 22 66 5 15 27 81 4 15 6 15 6 5 5 25 2 1 7 35 6 4 24 4 24 7 1 7 1 7 8 1 8 1 8 Ttal 28 471 186 227 466 698 Table 2: Number faccidents and affected peple fr multiple bunals (cmpletely bmed peple fund Withut visible parts frm 197 t 1999). Turs Off-piste Ttal Number f Cases affected peple Cases affected peple Cases affected peple burials % % % % % % 1 65.36 38.85 83.87 68.72 72.75 48.57 2 17.5 2.81 12.37 2.26 15.45 2.63 3 7.86 14.1 2.69 6.61 5.79 11.6 4 5.36 12.74.. 3.22 8.6 5 1.79 5.31 1.8 4.41 1.5 5.1 6 1.43 5.1...86 3.44 7.36 1.49...21 1. 8.36 1.7...21 1.15 Ttal 1 1 1 1 1 1 Table 3: Percentage f accidents and affected peple fr multiple bunals (cmpletely buned peple fund withut visible parts frm 197 t 1999). If peple get buried cmpletely by an avalanche the chance is quite high that mre than ne persn is buried. 2.3 Avalanche size ffatal and nn-fatal accidents Between fatal and nn-fatal recreatinal accidents there are significant differences cncerning verall length and width fthe avalanches (Figure 3 and 4). The smaller the avalanche the bigger is the chance t ski ut f it. The prbability fr lwer burial depth is als higher. As described in previus chapter the burial depth fr survival peple is much lwer (7 cm) than in fatal cases (12 cm). The mean fracture depth is the same fr fatal and nn-fatal avalanches (Figure 5), namely arund 5 cm. This crrespnds t the depth at which a persn can induce easily a failure in a week layer (Schweizer, 1998). Cmparisns between ski tur and ff-piste (Schweizer et ai., 2) and als between the different avalanche danger degrees (Harvey, 22) did neither shw any significant differences in the mean fracture depth. I---!------=====:--- fatal nn-fatal Figure 3: Overall avalanche length f fatal and nnfatal recreatinal accidents frm 197 t 1999. 451
Internatinal Snw Science Wrkshp (22: Pentictn, B.C.) 35 " 25. 2 ' 15.8 1 z 5 1% 6% 8 nt-fatal Figure 4: Avalanche width f fatal and nn-fatal recreatinal accidents frm 197 t 1999. "' N Existing traces Figure 6: Existing traces in an avalanche slpe befre the accident. Ttal number f cases with accrding data: 517. On ski turs the amunt f accidents which ccurred n descents is slightly higher than n ascents. But the number f caught peple and the number f fatalities is clearly lwer (Figure 7 + 8). 5% mre peple get caught while ascending. Avalanche accidents with several fatalities happen very ften while ascending. 2.4 Triggering fatal nn fatal Figure 5: Mean fracture depth f fatal and nn-fatal recreatinal accidents frm 197 t 1999. 2/3 f all recreatinal avalanche accidents in Switzerland get triggered n turs, 1/3 ff-piste. In 6 % f the cases the avalanche slpe had n visible traces befre the accident and the avalanche gt triggered by the first persn skiing the slpe. In 3 % the slpe was skied already n the same day f the accident. Only 1 % f all accidents ccurred while ascending in an existing track (Figure 6). en 'E 4 - '13 u<ti ' 2 Q5. 1 E :::J Z Turing descent <1> 274 Kind f activity Turing ascent Figure 7: Number f accidents n ski turs n descents and ascents. Ttal number f cases with accrding data: 574. 452
Rescue and Survival 1 944 8 a. 6 ' 4. E::J Z 2.5 Damage t peple 8j--------------:;::::=====:::l N g 2 Descent Ascent Kind f activity Figure 8: Nwnber f caught and dead peple n ascents and descents. Number f peple with accrding data: caught: 1577; dead: 455. Mstly males between 17 and 3 years are invlved in avalanche accidents (Figure 9). In 3 % fthe cases the size fthe grup causing the accident cnsists f 2 peple, t 16 % f 3 peple and t after all 12 % f lners. 8 % fcmpletely buried lners died in the avalanche. Fr all cmpletely buried peple this part is 5 % (Tschirky et al.,2). Therefre the chance t survive in an avalanche is significantly lwer fr lners, ifttally buried. Turing Off-piste Ttal Snw parameters: New snw (HN; cm) Ttal snw depth (HS; cm) Temperature fsnw (TS; c) Weather parameters: Wind velcity (kts) Air temperature (TA; c) Days with fur r mre recreatinal avalanche accidents (24 days with ttally 128 accidents) were defined as avalanche days and were cmpared with bservatin statins. Therefre fr each accident site the measured data f the nearest bserver was cnsidered in the analysis. Fr the statistical evaluatin the snw and weather data were transfrmed t fllwing nine parameters which were used fr a cluster analysis: Swn fnew snw 7 days back (sum HN) Swn fnew snw 3 days back (3d HN) Swn f new snw f the first 4 days ut f 7 (4dHN) Percentage f ttal snw depth t lng time mean (% HS) Air temperature difference frm accident day t day befre (TA diff6/7) Snw temperature difference frm accident day t day befre (TS diff6/7) Swn f wind velcities 7 days back (sum wind; ne measurement per day) Sum fwind velcities 3 days back (3d wind); ne measurement per day Swn f wind velcities f the first 4 days ut f7 (4d wind); ne measurement per day. '-r-----.---.-------,----.---.----.-------,-----.--,j 1 2 3 4 5 6 7 8 9 Figure 9: Age f caught peple in recreatinal avalanche accidents n turs and ff-piste. Ttal number fpeple cnsidered: 15. 3. Cmparisn f avalanche days with snw and weatherparameters The avalanche warning service f the SFISAR is based n an avalanche and snw bserver netwrk cvering the whle Swiss alpine area. Abut 8 human bservers wh measure fllwing snw and weather parameters which were used fr cmparisn in this study: The percentage f ttal snw depth t the lng time mean (% HS) is an indicatr fr a brief evaluatin f the quality f the snw cver (e.g. a lw percentage indicates unusually little amunt f snw and therefre a higher prbability fr snw metamrphism and frmatin fdepth har). Fr each avalanche day the mean f the nine parameters abve f each accident was calculated. With a cluster analysis (k-means methd) 5 clusters culd be evaluated with similar avalanche days cncerning the nine parameters. In Figure 1 t 13 the distributin f the relevant parameters fr avalanche frmatin (new snw, temperature, relative ttal snw depth and wind) is displayed fr each cluster. 453
Internatinal Snw Science Wrkshp (22: Pentictn, B.C.) 7 6 i 4 f s 3-2 iii 1 Figure 1: Distributin f sum f new snw 7 days back fr each cluster (sum HN). 15!!.. - 5 _.- en 1 I.. 5- '- 2 3 4 5 Cluster Figure 11: Distributin f percentage f ttal snw depth t lng time mean (% HS)... -5-1 iii I) -'_.!! 2 3 4 5 Cluster Figure 12: Distributin f air temperature difference frm accident day t day befre (TA diff617). 2 3 4 5 Cluster Figure 13: Distributin f Sum f wind velcities 7 days back (sum wind). The fllwing temperatures referring t 2 m abve see level ccurred in the mrning fr each cluster: Cluster 1: -6 C Cluster 2: -12 C Cluster 3: -1 C Cluster 4: -9 C Cluster 5: -1 C The resulting 5 clusters can be described as fllwing: - Cluster 1: Temperature rise and sparse new snw 23 Accidents spread ver 4 days were put int this cluster. The mean 3d HN was nly 1 cm. It's the nly grup with a rise f air and snw temperature n the accident day. In 2 cases the predicted avalanche danger degree was "lw". - Cluster 2: Lts f new snw (specially 3d HN) and cld temperatures 25 accidents spread ver 5 days were put int cluster 2. The largest amunt f new snw fell during the 3 days befre the accident (n an average f 36 cm). The temperatures were cld (all mean temperatures belw -1 C at 2 m abve see level). - Cluster 3: relatively weak snw cver, mderate amunt f new snw and winds, cld temperatures. 45 accidents spread ver 7 days were put int this cluster. As in cluster 5 the percentage f ttal snw depth is als lw and indicates a bad stratigraphy in the snw cver. 454
- Cluster 4: Strng winds On 3 days 17 accidents ccurred in this cluster. The mst characteristic parameter in this cluster are the strng winds which caused dangerus wind drifted snw. - Cluster 5: catastrphe cluster 18 accidents spread ver 4 days were put int this cluster. The largest amunt f new snw was measured and the winds were medium. The mean snw cver is indicated t be the weakest f all clusters. In this cluster many relevant factrs fr avalanche frmatin were given. Frm 197 t 1999 "nly" n 24 days mre than 3 recreatinal accidents happened in Switzerland. The weather and snw situatins befre the accidents were nt extrardinary n these days but ther circumstances matched t achieve an avalanche day. Fr example n mst f these days the weather was beautifully sunny and they ften fell n Sundays r hlidays (12 f 24 days were Sundays). On such days many peple are skiing in back cuntry terrain and therefre the ptential fr a damage t peple increases. On mst avalanche days the danger degree "cnsiderable" was predicted by the Swiss avalanche warning service. On the 1. January 1997 (cluster 1) tw accidents ccurred at danger degree "lw" and tw mre at level "mderate". Except fr the temperatures all ther snw and weather parameter were rather psitive (Sum HN = 13 em; 3d sum HN = 8 em; % HS = 136 % (the snw pack was als judged as gd by the frecasters); 3d wind = 9 knts). The temperatures n this day rse by abut 8. On the day befre -12 C was measured at 2 m, n the accident day -4 c. The nly indicatr fr an increase f the avalanche danger was the temperature rise, which is ne f the mst difficult parameter fr a frecaster and practitiner t estimate what the cnsequences are. 4. Cnclusin Fatal avalanches are significant larger than nn fatal accidents. But the mean fracture depth is the same fr all accidents and crrespnds t the ideal depth t induce a failure in a weak layer. Althugh the stress induced in the snw cver is larger n descents than n ascents, mst peple n turs get caught while ascending. The reasn is, that ascending peple ften walk quite clse t each ther and therefre mre than ne persn is situated in a dangerus zne. Furthermre n ascents yu are Rescue and Survival slwer t react and try t get ut f the avalanche. The fact that nearly 4 % f recreatinal accidents ccur in slpes with existing tracks in it shws, that reliance n existent tracks is nt the best advise. Hwever it has t be taken int cnsideratin that frm many accidents there is n infrmatin abut the existence f traces in the slpe befre the accident happened. In existing ascending tracks hardly any accidents happen. The chance f ttally buried peple t be invlved in multiple burials is higher than expected (61 % fr turs and 31 % fr ff-piste) The cmbinatin f avalanche frming parameters with sunny weather and hlidays results ften in an avalanche day. Mstly a reasnable amunt f new snw (3-6 em in 7 days) was required fr an avalanche day. But in 2 % f the days there was neither a significant amunt f new snw, nr strng winds, but an increase f temperature. 5. References Harvey, S. 22. Skifahrerlawinen und Lawinenbulletin: Ein Vergleich. Die Alpen 4/22. Schweizer Alpen Club (SAC). Jamiesn, J. B. and Geldsetzer, T. 1996. Avalanche accidents in Canada - Vl. 4: 1984-1996. Canadian Avalanche Assciatin, Revelstck BC, Canada Schweizer, J. 1998. Cntributin n the rle f deficit znes r imperfectins in dry snw slab avalanche release..prceedings Internatinal Snw Science Wrkshp, Sunriver Oregn, USA, 1998. Schweizer, J. and Liitschg, M. 2. Characteristics f human triggered avalanches. Prceedings Internatinal Snw Science Wrkshp, Blue Sky MT, USA, 1-6 Octber 2. Signrell, C. 21. Skifahrerlawinenunfalle in den Schweizer Alpen. Eine Auswertung der letzten 3 Jahre. Diplmarbeit Eidg. Institut fur Schneeund Lawinenfrschung, SLF, Davs. Tschirky, F., Brabec, B. and Kern, M. 2. Avalanche rescue systems in Switzerland: experience and limitatins. Prceedings Internatinal Snw Science Wrkshp, Blue Sky MT, USA, 1-6 Octber 2. 455