Ken Roberts -- Ski Backcountry

see also:  other notes on Skiing this year

toward a framework for managing crevasse risk on skis


what's here in detail

on the more on crevasse risk page:

some specific issues:

  • good days versus bad days

  • roping while climbing up versus skiing down

  • hiring a guide

Here's some tentative ideas proposed by an amateur with no significant expertise. Likely it has some mistakes in it, perhaps some pretty serious or dangerous mistakes -- so use these ideas as stimuli to raise questions and think more for yourself and ask other people who know better.

My purpose in putting them out in public is that I'm hoping it will stimulate somebody with real expertise and much more experience to write something better -- which might point out and fix some of the errors here - (or even better, make this page just irrelevant, so I could just remove the content and replace it all with a link to the true expert page).

These ideas based on experiences skiing and climbing on crevassed glaciers, mostly in France and Switzerland (and a bit in the Washington Cascades), and on some reading some English-language books and trip reports about it.

My focus here is on the Alps of Europe -- I don't know how these ideas might (not) apply to other regions (e.g. Alaska, Canada, Himalayas).

What makes Europe special:

  • lots of ski tourers (on some routes),

  • lots of guidebooks and reports,

  • lots of weather and snow condition info,

  • lots of historical knowledge,

  • limited range of altitudes.

straightforward procedures

It's reasonable to want to find a concise description of straightforward procedures for managing crevasse risk.

This page does not offer this.

This page instead offers complicated messy analyses of crevasse risks and strategies for skiers. It gives lots of alternatives, without attempting to say which one you should believe or use.

how to help us toward something better

  • post links and titles of helpful English-language resources on public skiing forums (e.g., TelemarkTalk, TGR)

  • post links and titles of helpful French- or German-language resources on public skiing forums.

resources in Italian or other languages might also be helpful, but French and German is what I know some ways for me to understand or get translated -- other non-English languages help me only if you can find someone else who can summarize or translate.

  • post other ideas about principles and strategies for skiers to public skiing forums (e.g., TelemarkTalk, TGR)

  • post corrections to these ideas to public skiing forums.

how to be safer

If you really want to be safe from crevasse falls, don't travel anywhere near any place which might have a crevasse.

If you want be be "reasonably safe" from crevasse-fall risk while skiing on a glacier, do it when the snowpack and weather history are favorable (more on this below), learn some basic dos + don'ts of glacier skiing tactics, and hire an expert professional mountain guide who lives near and skis a lot on the glaciers you want to ski.


My basic ideas here are that:

(a) skiers should handle crevasse risk more like they manage avalanche risk: like assess situation by "red - yellow - green" or "1 - 5" rating scales, and understand what locations have special risks. There are some interesting similarities in the physics and human experience of these two hazards -- but crevasse risk has two advantages: the physics tends to be less complicated, and being on skis helps.

But avalanche risk tends to have more knowledge -- because it has been studied much more -- because avalanches impact roads and houses and villages -- and sometimes skiers on groomed marked trails at ski resorts who paid for lift tickets -- so there are clear political and financial motivations to learn to do better with avalanche risk.

(b) skiers typically should handle crevasse risk differently from non-skiing climbers.

But most of the existing books + articles + web pages mainly take a climber's perspective. So more work is needed to offer a strategy better suited for skiers. This page makes a non-expert attempt to dig into some of that work.

key new ideas here

Here's list of links to sections with ideas that I haven't seen before in other books or websites.

(which means that likely a significant percentage of the ideas are mistaken, and some might even be dangerous)

Principles of crevasse risk: I think most of those principles are not new, but perhaps it's the first time they've put together in one list.

complexity and variations

Likely smart experienced skiers who do different kinds of tours in different regions are going to develop different focus in their strategies: like the eastern Alps tend to have narrower crevasses, and the high-altitude tours in the western Alps tend to have wider crevasses, and managing the probabilities and hazards and consequences of big wide crevasses is likely to be different than for narrow crevasses. Tourers who try to ski glaciers mostly in the winter are going to have a different focus than skiers who go out in late spring. Risks and techniques for gentler slopes (with gentler easier descending and straighter climbing) might be different than on steeper slopes. Regions with "maritime" snow are different from regions with a "continental" snowpack.

skiers versus climbers

I think most of what's been written about this in English tends to focus on non-skiers traveling on glaciers, but I think there are often some important differences in the goals and situation of skiers from non-skiers -- so I think strategies and principles focused specifically for skiers deserves a separate discussion.

Key differences:

  • non-skiing climbers are typically on glaciers in seasons where the crevasse bridges are usually weakening (because of melting and lack of substantial new snowfall), so they cannot feel confident that the bridges will hold them if they held the climbers who walked the same track a week ago. They cannot necessarily feel confident that the bridges that held them in the morning will hold them following exactly the same steps in the opposite direction in the afternoon.

On many of the (remaining) glaciers of Europe, in many years there are some weeks where ski tourers can feel confident that the bridges are usually strenthening, or at least stable -- unless there is some special circumstance: like fresh snow together with wind, or sudden warming.

  • climbers are often expecting to want to "rope together" for protection on other non-glacier sections of their climb.  Or at least they frequently do climbs (at least in "practice" areas) that use the rope for protection. They are accustomed to moving while attached to the rope for protection, and to managing the rope while somebody else if moving. And accustomed to the kinds of rope work which might be necessary to carry out a rescue.

Most skiers rarely practice traveling roped together, even for climbing up -- and never have practiced roped together while skiing down. Backcountry skiers who do not do technical climbing in the summer typically feel very little reason to practice basic climbing knots.

  • roping together is much more hassle and risk for skiers than for non-skiing climbers.

Obviously so for hassle for skiing down, because it's rather difficult for everyone in the team to descend at the same speed and in the same direction. Very difficult to manage the rope while skiing to compensate for inevitable differences in speed and direction.

Also more hassle while climbing up on skis, because on moderately steep terrain skiers tend to climb on zig-zag path with kick-turns, so their relative direction of travel is changing.

More risk, because if one skier breaks through into a crevasse, it's harder for others (especially if above) to arrest against the pull with their skis on.

  • therefore, climbers typically are ready and willing to rope together when they think there's any small risk of crevasse fall.

Skiers tend to try to avoid roping together. If they think a tour might require roping together, they tend to just choose a different tour -- or postpone considering this tour until a different week (or a different year) which they hope might be more favorable.

If they think a tour might call fo roping together for skiing down, they surely will choose a different tour.

  • therefore, climbers are less concerned about assessing the degree of risk, and more focused on carrying the right equipment and learning good techniques for rope management and rescue.

Skiers want to get really good at being able to assess which routes on which days have no significant crevasse risk, so they don't have to be bothered with hassle of rope management or the risks of arrest and rescue after a fall.

  • books about crevasse risk and strategy tend to have lots of information about equipment and techniques for rope management and rescue, because there's lots of straightforward specific things to be said about them -- and interesting diagrams to draw.

While risk assessment is tricky and tends to lack definitive tests or measurements -- because relevant data is not easy to obtain for specific crevasse zones, and because crevasse risk just hasn't been studied as much as avalanche risk.

  • therefore books and articles and web pages about crevasse risk are much more helpful to climbers than to skiers.

so there's lots of work remaining to be done on strategies for skiers.

how skiers get hurt or die in crevasses in Europe

Here's some theories . . . some based on trip reports . . .

  • They have no clue about crevasse risk. They just want to ski.

  • They do not change their trip plans based on whether this season has high or normal or low snowpack at high elevations. They have no fall-back plan for if their trip week (scheduled three months ago) now happens to have inherently dangerous crevasse-fall conditions.

  • They ski routes in mid-February which usually ought to be left until April.

  • They try to ski untracked powder on crevassed glaciers without a professional local guide - (instead of waiting a few days for snow-bridges to consolidate, and in the meantime looking for powder on non-crevassed slopes).

  • They don't know which routes are "plugged" in spring season in a normal snow year versus which ones are "plugged" only in a big snow year - (and they don't make much effort to find out).

  • or . . . sometimes they're just very unlucky, even after doing lots of things right. Just like sometimes careful skiers get caught in a freak avalanche.

alternate strategies list

Here's some different possible strategies which skiers might try to manage crevasse risk:

(z) Never ski (or climb or travel) on crevassed sections of glaciers at all.

(a) follow normal ski travel routes at normal times of the year.

(b) If other people have skied it, then it's probably OK, so just wait.

(c) Rope together at all times on any section of glacier known to have underlying crevasses: both climbing up and skiing down.

(d) Always rope together for climbing up any section of glacier known to have underlying crevasses, but normally not for skiing down unless there's a reason to think it's more risky than usual.

(e) Before starting a tour or entering a crevassed section, make an assessment of the crevasse risk based on weather + snow history and observations.

(f) Perform specific tests out on the snow while climbing up thru crevassed section, to assess if the risk seems low enough to justify skiing back down thru it unroped.

(g) Hire a qualified mountain guide.

pros + cons of different strategies

(z) Never ski (or climb or travel) on crevassed sections of glaciers at all

This assumes that you know (from recent summer photos? recent reliable trip reports?) which areas that you're considering skiing on have no crevasses (visible or hidden) all year long.

Note that large hidden (or visible) cracks and holes can form in snow which is not on a glacier and which are not defined as "crevasses" -- but are still capable of causing death or serious injury if you encounter them. Sometimes these are called "moats", "glide cracks", "wind holes", or "tree wells", or a "hidden creekbed". Some snowfields have no crevasses other than a single "bergshrund" or "rimaye" - and some people might not call that a "crevasse", but it's something you'd want to know about and be careful of.

A different way to take the motivation of this strategy is to ski only on either year-round uncrevassed areas -- or sections whose crevasses are currently so well-bridged ("plugged") that there is no way for a skier (or even multiple skiers gathered in one spot, or a single person skiing aggressively) to break through a bridge.

pro:  simpler than other strategies, safer than other strategies.

con:  miss out on some great ski tours.

(a) follow normal ski travel routes at normal times of the year

pro: Little understanding or data required -- except need to find resources (e.g. guidebooks) which indicate which glacier routes are especially risky (e.g. "tormented") even in "normal" years, and which ones are not "normally" risky for skiers at "normal" times.  For many glaciers in Europe, this knowledge is available (at least for those who can utilize resources in French and/or German)

con: Works except when it doesn't. Reasons why it doesn't: (1) some years are inherently dangerous; (2) some snow + weather histories are inherently dangerous; (3) the glacier changes over several years: a route that was once usually safe might now often be dangerous.

Variation: Follow only usually-safer-than-most ski travel only in mid-spring only in a big snow year.

And when it's any more risky than that, instead travel in springtime to a region with very little crevasse risk (e.g. the Sierra mountains of California USA).

(b) If other people have skied it, then it's probably OK

pro:  There is a definite grain of truth to this:  Other people skiing it (if you know they made it all the way thru that section of the glacier without punching into a crevasse) -- that counts as evidence that the crevasse bridges are stable. It's some of the most specific evidence that you can obtain.

con:  Big problem is that the situation when crevasse risk is high due to new snow + high winds + without many consolidation cycles is also when the strengths of crevasse bridges are most variable. So the strength of the bridge where the previous track went is not such good evidence of its strength 3 meters away.

Or when there's sudden warming, so the bridges are tending to weaken.

(c) Rope together at all times, for skiing down as well as climbing up

pro:  A simple rule, which generally results in reduced risk. Climbers tend to like this rule, also people who have spent time staring at detailed aerial photos taken around September of glacier ski routes.

con:  Even when roped, punching through into a crevasse can be harmful (especially when skiing downhill, where precise rope control is more difficult) and it can be a complicated mess to get someone up out of a big hidden crevasse even if they were roped.

There are situations on skis where the risk of punching into a crevasse is very very low, so the reduction in risk is very small, while the reduction in speed of travel is significant (and the reduction of the fun of skiing). In some situations slower travel time is itself a increased risk (sometimes an increase in crevasse-fall risk, if it's a hot day)

con: I've never heard of any skiers in the last 50 years who followed this rule on any tour. Skiers will simply choose a different tour, rather than follow this strategy.

(d) Rope together climbing up, but not normally when skiing down

(d) Always rope together for climbing up any section of glacier known to have underlying crevasses, but normally not for skiing down unless there's a reason to think it's more risky than usual.

pro:  Simple rule, which generally results in a small reduction of risk.

con:  But the crevasse risks while skiing back down are significantly greater, so it's not solving the bigger problem.

pro:  In some special situations, the crevasse risk while skiing back down might not be greater -- perhaps situations with smaller crevasses on a gentle slope in cool weather, and with an appropriate downhill-skiing strategy ?

con: The temptation is to think that since we didn't punch through while climbing up, then likely it's OK to ski back down through that section.

This reasoning runs into these problems:

  • warmer air and sunshine tend to soften hidden bridges as the day progresses, especially in springtime.

  • skiing down will not exactly follow the track of climbing up: Big problem is that the situation when crevasse risk is high due to new snow + high winds + without many consolidation cycles is also when the strengths of crevasse bridges are most variable. So the strength of the bridge where the previous track went is not such good evidence of its strength 3 meters away.

My personal experience on long interesting descents is that it can be difficult to maintain the mental discipline of following the safest line -- get distracted by taking photos, relaxed after making it thru a tricky section, nothing bad has happened so far . . .

  • while climbing up, the leader had time to select the safest path, but skiing back down will not have time to select so carefully -- so the uptrack systematically understates the risk of the skier's downward paths.

The uptrack can be planned to aim across the width of many crevasses in a possible dangerous section, which the curved arc of a downhill turns includes many relative angles, including some more along the length of the crevasse - (skis aimed more along the length tends to be more likely to punch through the bridge than skis aimed roughly across the width).

  • the motions of downhill skiing tend to put more pressure on hidden bridges than uphill walking, at least in soft snow where most skiers tend to use some up + down motion of their upper body to help initiate turns and stops.

Though I doubt that the additional pressure at the finish of the downward phase of normal soft-snow ski turns adds more than 5% to the downward pressure already on the snow from weight of skiers body (and usually less). Perhaps jump-turns could add more. Surely taking air or drops adds lots more impact.

On the other hand the higher speed going downhill might get you across a collapsing bridge over smaller crevasse before you fell in -- but that doesn't work so well on steeper slopes where you're making tigher turns to control speed, so you're speed is lower between turns and your skis might tend to be aimed in the line of the crevasse (not across it) at the same time -- and in soft snow you tend to be finishing the down phase of your vertical upper-body motion, which is adding pressure beyond static body weight -- so you simultaneously lose the three advantages of ski speed and ski direction and static upper body.

Therefore if you believe that it's an important risk-reduction policy to rope together while climbing up, then (often?) it's even more important to rope together skiing down -- strategy (c).

My warning questions:

(1) If you think the crevasse risk is enough significant so you feel you need to reduce it during the uphill phase by roping together, then why are you attempting that tour on that day at all?

(2) If you're "fairly" confident it's safe enough to ski down, but want some confirmation out on the snow first, then do a serious testing procedure while roped together (see below), not just "climbing up".

My basic concern with this strategy is that it tends lead people to use a simple rule in order to avoid the complicated task of making an assessment of the risk level before starting the tour.

(e) Before starting a tour, make a "prior" assessment of risk level

(e) Before starting a tour or entering a crevassed section, make an assessment of the level of crevasse risk based on weather + snow history and observations, and on previously-known risky terrain features (e.g. from summer photos or from guidebook descriptions), and on recent trip reports from other ski tourers.

Comparing with avalanche risk . . .

most ski tourers make this pre-tour "prior" assessment the main focus of their risk management strategy -- unless they see obvious evidence while out on the tour which contradicts their prior expectations. Most ski tourers I know normally leave detailed on-snow testing procedures to professional forecasters who have the time and expertise to do a good job of it. But . . .

For crevasse risk, many ski tourers do not put much serious effort into making a "prior" assessment of level of risk for crevasse-fall hazard. I think there are some reasons:

(1) They don't know how to make such an assessment for themselves;

(2) There is no published "official forecast" for crevasse risk level;

(3) There are no standard definitions and formats to report observations about crevasse coverage in a trip report;

(4) Useful information such as a histogram of crevasse widths in a section of glacier (current as of the last year or two) are not readily available - (and it's not necessarily easy to estimate this histogram from recent photos).

For some hints about assessing the risk level, see under red light - yellow light - green light.

pro: It fits with the strategy which skiers are likely to use anyway once they get out there on the snow, including (a) not doing serious testing on the way up; and (b) not roping together on the way down.

con: For many glaciers and many situations, it's hard to get accurate relevant data. See above under "why many ski tourers don't do this assessment".

(f) Perform specific tests out on the snow

(f) Perform specific tests while climbing up thru crevassed section, to assess if the risk seems low enough to justify skiing back down thru it unroped.

Question:  How is this different from just climbing up while roped?

Answer:  The point of testing is to deliberately seek out and specifically test potentially risky stresses and locations and configurations while roped - (perhaps tailored to focus on potential problems identified from weather + snow history and on previously-known risky terrain features).

Approach to testing:

The idea would be to apply stresses and explore paths while climbing up which simulate the stresses and paths which might be applied when skiing back down. There's no way to simulate the added risk due to higher temperatures later in the day, so it makes sense to apply stresses of greater intensity than you would expect from downhill skiing.

Of course these tests might result in immediately punching through into a hidden crevasse, so it's crucial to perform the tests while appropriately using rope and full crevasse-rescue safety equipment (and warm clothing), and having frequently and recently practiced a full range of crevasse-rescue procedures.

Of course for testing like this, the party size must be larger than two persons - (since rescue with only one person outside on the surface while the other is down in the crevasses is inherently problematic). Also for a larger group of skiers, it's easier to justify the extra time and risk taken to perform a testing procedure.

Makes sense to me that you would only be performing these tests if you were already fairly confident from pre-tour "prior" analysis of weather and snow history and terrain features that it was likely that this section of the glacier would have low enough risk for you to want to ski downhill thru it unroped.

It is true that performing special testing procedures will increase the probability of punching down thru into a crevasse during the testing, and even with appropriate equipment and careful procedures while roped together this might result in injury or a time-consuming rescue procedure.  But . . . (1) how much worse it could have been if a similar fall had occurred while skiing down unroped; (2) If you have not convinced yourself from prior analysis of weather and snow history that crevasse-fall danger is likely very low, then why are you, full of hopes for fun downhill skiing, out there on that glacier section on that day at all?

Ideas for specific testing:

  • do some stomping or even jumping to apply greater stresses to the snow than normal uphill walking - (how much greater might depend on how "fun" or "aggressive" downhill skiing you expect by some members of your party, e.g. taking air off humps?)

  • since higher-risk situations have greater variability, need to test a larger sample number of locations.

  • since "fun" downhill skiing is less careful than uphill walking, might need to test a wider variety of kinds of terrain features - (how wide a variety might depend on how careless you expect some of your party).

  • since downhill skiing normally uses a wider path than uphill walking, makes sense to make multiple uptracks which "cut a wider swath" than the normal single-skier uptrack.

  • a procedure to try to address several of these ideas is for each skier of the party to climb on a separate uptrack, roughly a fixed sideways distance from the previous skier. (what length to choose for this sideways distance is a tricky judgement call). So the rope connecting the skiers runs diagonally relative to the overall direction of travel. This gets more complicated and tricky in sections steep enough to require switchbacks zig-zagging side-to-side across the slope. Of course in fresh snow this procedure is going to require a lot of extra work trail-breaking - (another reason to avoid crevassed sections soon after a substantial snowfall).

pro:  If done properly, it's highly specific, highly relevant, often not difficult to interpret.

con: (a) Takes a significant of time.

(b) Doesn't look "pretty" on the snow.

(c) Experienced glacier skiers often would think the extra time was unnecessary because the risk was so low.

(d) Experienced tourers who did not think the risk was so low might instead just choose a different tour.

(g) hire a qualified mountain guide

Hiring a guide usually makes skiing on a glacier safer, but it does not make if fully "safe".

Deciding to ski a tour + day with several "red light" factors (see below) and then hiring a guide does not make it into a "green light" day.

Mountain guides are accustomed to taking risks, and they are accustomed to clients who are ready to take some risks -- and if they want to make money they're willing to take clients out on days which are riskier than other days.

So when you approach a mountain guide or guide service, most of them do not consider it their role to tell you things like: "This glacier tour would be safer if you waited two days" or "if you waited until next month" -- or "really safer if you waited until next year".

They do not consider it their to show you some August aerial photos of the glacier tour you want and say "Look:  crevasses that could injure or kill you underneath almost everywhere you'll be skiing".

Once you're out there skiing, the guide can't stop you if your emotions of joy or fear or pride lead you to do something dangerously stupid. Unlike alpine climbing with a guide, you're not roped to the guide, so you have freedom.

Some guides are more risk-averse than others, and it's hard to know in advance whether your particular guide is is a good match for your own desired level of risk-aversion (or your own level of risk-seeking). If you go with a guided group, you tend to get dragged along with some sort of default presumption about the risk preferences of "the group" - (not necessarily bad, since you might not know your own glacier skiing risk preferences very well anyway).

So even with hiring a guide, there's lots of room for "tuning" your risk at the macro level and the micro level -- by choosing what season and day to try to ski on a glacier, and by reading about (and practicing) how to handle specific tactical moves and contingencies while you're out there skiing.

Also, even the  best guides sometimes miss seeing things or forget to say things -- so asking questions can be helpful for keeping them alert.

Local knowledge: Some guides know the have lots more experience over several years in different seasons of the year, and the current season snowpack history, with the specific glaciers you want to ski on. One of the findings of the analysis of glacier skiing risk here is that specific knowledge really helps - (perhaps more so than for alpine climbing). So while a good non-local guide usually helps significantly with managing risk, there's another level of sophistication which is missing.

Bottom line:

A guide cannot make glacier skiing fully safe because: (1) Traveling on a glacier with crevasses hidden underneath the snow is fundamentally dangerous and out of control of the guide. (2) Your own skiing is out of control of the guide. (3) You're all assuming that you're not using the possibly(?) safer method of skiing down very slowly and carefully while roped.
Therefore, hiring a guide with specific local experience usually helps reduce risk of crevasse fall significantly, but there's still a large difference you can make by your macro + micro decisions.


  • there's basically two kinds of crevasse falls: "punching through" a section of a bridge over a crevasse, and "collapsing" a whole side of the bridge over a wide crevasse (like collapsing a cornice).

  • skis versus boots:  traveling on skis tends to have a significantly lower chance of "punching through" a locally thinner section of a crevasse bridge, or over a non-wide crevasse, than traveling on boots.

because it spreads the skier's weight and impact over a larger area.

So risk management strategies for non-skiing climbers and hikers tend not to be helpful for skiers.

So don't take your skis off. If need to take skins off or onto skis, take off only one ski at a time.

This could be a reason to get good at skinning up and making kick-turns climbing up steep slopes, instead of booting up and carrying skis.

Snowboarders may find it difficult to avoid traveling on boots only on flat sections of the glacier.  A snowboarder on a famous downhill route died this way in 2008 -- they ran out of glide on a flat section of the glacier, so they took off their snowboard and started walking in their boots to reach a point where they could start sliding downhill again, and then they punched through and hit something down in the crevasse and died.

But skis do not help as much with the danger of "collapsing" a whole side of the bridge over a wide crevasse.

  • crossing a (non-wide) crevasse on skis across its width tends to be safer than roughly along its length.

So it's useful to know the general direction of crevasses on a glacier (e.g. from maps or summer photos) for planning specific route and for identifying special danger situations. Skiing downhill usually has curved turns with a wide arc, so it tends to cross crevasse bridges at many different angles, which typically includes some more along its length -- another reason why skiing down tends to be more risky than climbing up.

  • getting more than one skier together in one spot is much more dangerous.

because it focuses more weight on a single part of a snow bridge. Also because it might be enough weight to collapse one side of a big bridge over a very wide crevasse, and falling into a big crevasse is very bad idea -- and having multiple persons falling into the same big crevasse is a very bad problem.

So find a spot which you have good reason for knowing that it's not on a snow bridge.

Another strategy sometimes tried is that the first person who goes to the spot not known, and probes into the snow around it a lot (like with an avalanche probe pole, since an ice axe shaft can't probe very deep) checking for hidden weaknesses. But in winter if the snow fallen so far in the current season has not been through some significant warming-cooling cycles, even a very thick layer of snow could collapse under the weight of multiple skiers -- so probing would not detect that situation.

Emotions tend to bring skiers together. Scary scenario:  One skier punches thru a thin spot in the center of bridge covering a wide crevasse. Three other skiers come to the hole to help, and then the whole bridge collapses.

Non-emotions tend to bring skiers together. Humans just like getting together when they stop moving. I think it takes practice and discipline to keep remembering to stay apart each time you stop.

Simplest policy is just don't come together until you're off the glacier. When you need to talk on the glacier, still stay apart, and talk loudly.

Carrying 2-way radios could be important, especially on windy days.

  • following closely a previous ski track (which did not punch through) is often less risky than taking a different route farther away

(except that sometimes following a previous track is more risky)

If the previous track was say at least a day before, and the snow was not too cold, then the track of the previous skiing party could actually strengthen the snow: by "sintering". But if it was a cloudy or warm night, or temperatures are warmer than the day the previous track was made, then it might be weaker.

If it's still the same day, sometimes it might be safer to ski close to but not actually on the previous track: Because the previous track might have temporarily weakened it's particular spot over a bridge -- but on the other hand the spot one meter to its side might have been weaker to start with, and so could collapse under your weight even without temporary weakening -- which shows that getting too clever doesn't always help.

  • cornices -- snow around a wide crevasse can be like two cornices, one on each side edge, overhanging a giant empty space below. Just like on a ridge you need to be very careful about going toward the edge, same with going near to the edge of a crevasse in a section of a glacier which might have wide crevasses -- a long section of the side of the crevasse can collapse and fall down in.

Just like on a ridge where the cornice sometimes overhangs much farther than you might ever guess and can break off much farther from the edge than you might ever guess -- same with the snow around wide crevasses.

Somtimes two skiers might be 8 meters / 25 feet apart from each other, but standing on the same overhanging lip on the same side of a crevasse much wider underneath than they could have guessed -- and it's just like putting double-weight loading a cornice on a ridge.

Another reason why it's important to know the direction of the crevasses.

  • variations in the thickness of bridge sections often do not get "filled in" or "smoothed out" much during the season, once the bridges have fully formed.

That's because a major force in forming crevasses is wind along the surface of the snow, so the top surface soon tends to smooth out, and then the variations are mostly in the bottom of the snow layer. So if the snow near the side of a wide crevasse is 6 feet thick, but the center of the bridge is only 1 foot thick -- then after 3 more feet of snow have fallen, the snow near the side is 9 feet thick and the center is only 4 feet thick. The difference is still 5 feet. Then in late spring early summer, after 3.5 feet of snow melt, the sides are 5.5 feet thick and the center is only 0.5 foot thick.

So suppose the snow surface looked smooth in February, but there were some spots that were actually rather thin. In April the snow surface still looks smooth, maybe smoother. But if it's a low snow year and the April snowpack is only 1 foot deeper than in February, then the thin spots are likely only 1 foot thicker than they were in February.

  • aggressive skiing is usually more dangerous.

because it puts higher impact forces on snow bridges. Especially drops and jumps.

My rough calculation is that taking just a 1 meter drop or air typically has at least 5 times the impact of the skier's body weight on the snow.

Even a sudden sharp turn or stop that includes substantial up + down motion of the upper body adds some impact -- though likely not more than 5% added to the pressure of the skier's body weight already on the snow. (Ski turns with no vertical upper body motion do not add impact or pressure on the snow).

  • falling into a bigger wider crevasses is usually more dangerous than falling into a smaller narrower crevasse.

With a small crevasse, often you don't fall very far before getting "wedged". If not injured (or stuck), can sometimes climb out with "stemming" moves (easier with crampons). With a big wide crevasse, can fall a long ways before stopping, hit hard, major injury (or instant death). Getting out usually requires a rope and other specialized equipment (and skill in using them) -- especially tricky with a party of two (with only one person outside the crevasse) -- and for a solo traveler escape is often not possible (even if uninjured in the initial fall).

Therefore find out in advance which sections have wider crevasses, and be much more careful about skiing (roped or unroped) in those sections.

And do not move close together on a section which might have bigger wide crevasses -- because the combined weight of two or more skiers might collapse the bridge - (and having multiple skiers fallen down inside at the same time is a much bigger problem).

If the crevasses are small enough, the slope gentle or flat so that the skier can move at significant speed without using significant down-force for turning or slowing, and the skier knows the direction of the crevasses, and it is possible for the skier to consistently ski roughly perpendicular to the crevasses --  then the higher speed of downhill-skiing might carry the skier across to the other side even though the snow bridge over it partly collapses.

  • variability: Situations with higher risk tend to also have higher variability by location.

The situation when crevasse risk is high due to new snow + high winds + without many consolidation cycles is also when the strengths of crevasse bridges are most variable. Because wind distributes snow in variable ways, even then there's no obvious reason, and crevasses have variations in width, so the strength of newly formed bridges varies widely.

The situation when crevasse risk is high due to warm temperatures melting ("wasting") old snow bridges also has wide variations in which bridges are stronger or weaker. Because there was often wide variations in thickness and strength of crevasses when they formed, because there are variations in what happens to bridges as the glacier flows, and because in the warmer "wasting" season there are variations in how much sun different parts of the glacier receives and what percentage of sunlight is absorbed (versus reflected).

So in higher-risk situations, testing the underlying snow in one location does not reliably predict stability in a nearby location. The strength of the bridge where the previous track went is not such good evidence of its strength 3 meters away.

  • Knowing the physics of crevasse formation is often not helpful in deciding whether to ski unroped on a section.

Because it's not enough to know where more or bigger crevasses are likely to form, more important to know where crevasses are likely to be hidden. A glacier section with wider crevasses might sometimes be safer, if many of them are not bridged, because they're visible and so give valuable information about where bridges are and how thick, and likely alignments of hidden crevasses.

Because big glaciers in Europe often have substantial crevasses in sections which are thought relatively less probable (e.g. on wide sections of fairly uniform slope).

Because it's one thing to use general guidelines for selecting which section of the glacier to travel on when roped, but once a sequence of sections has been selected, the decision to travel unroped (rather than roped) on any of those sections is a different kind of choice -- usually made to increase fun -- so a higher degree of evidence is called for.

It's not good enough to say "physics suggests that crevesses are less likely in this section, so it's OK for us to unrope here". If want to have the extra fun of skiing unroped, need to do some extra work in advance to gather evidence: e.g. summer photos which show where crevasses actually are, or guidebook suggestions, or recent trip reports. (Especially in a highly-populated and traveled region like the European Alps where better information about glacier conditions and ski routes should be available.)

  • Change: The world of glaciers keeps changing. Some glacier sections which ten years ago were fairly safe in springtime in a normal snow year are now not so safe. Some sections which were usually not safe 10 years ago are now not glaciers at all.

  • Complicated: Especially around wide crevasses and icefalls and seracs, the physics can get pretty complicated. Strange local wind patterns, strange sequences of temperature and precipitation, etc. There's an inherent risk with unknown probability of a bad fall into a crevasse happening, despite every careful effort to understand the situation and choose the best route.

  • Having a recent summer photograph is rather helpful in analyzing the risks of crevasse fall - (especially if some numerical parameters are known so that the absolute width of crevasses in different sections of the glacier can be reliably estimated.

If absolute size of width can be reliably estimated, than can assess which sections have crevasses mostly narrow enough to be possibly climbed out of without a rope, or narrow enough so that an unroped fall might be hoped not to result in immediate serious injury from hitting hard at the end of the fall. (Also such narrower crevasses are more quickly and easily becomme strongly bridged).

Even if absolute size of width cannot be estimated, knowing relative sizes of crevasses can also be helpful, because after identifying which sections of the glacier have wide crevasses, which have narrower crevasses, and which have both -- then (a) if some of the wider crevasses are uncovered, then can suspect that some of the narrower crevasses are not strongly-enough bridged; or (b) is all of the wider crevasses in the summer photograph are now covered, then it is more likely that most or all of the narrower crevasses are more likely to strongly bridged. (Especially if multiple snowstorms have come and gone since the wider crevasses were covered).

  • Terrain features

This is covered lots in book and articles for climbers on glaciers (so read those for much more detail). Generally, more and bigger crevasses tend to form in sections where the flow of the glacier is getting "stretched". Obvious places are: on the outside of a curve, around a rock island, over a "convexity" (where the slope going down the hill transitions from gentler to steeper). Crevasses are less likely to form in a section of "concavity" (where the slope going down the hill transitions from steeper to gentler), and less likely in gullies (which also tend to hold more snow which could bridge them -- but more likely to be in shade which slows the strengthening of a bridge -- and if exposed to avalanche or serac-fall from above, could be a trap).

Some problems with using this knowledge: (a) the big danger for skiers is hidden crevasses, which might not be in the same section as the bigger crevasses; (b) Plenty of potential lethal crevasses form in other sections of the glacier; (c) larger crevasses can form near a large rock hidden under the surface. (See above under "Knowing the physics ... is often not helpful")

  • Potential bridges in winter and early spring tend to "resolve" quicker in warmth and sunshine than in cold and shade.

By "resolve" I mean the snow-covering either collapses so the crevasse becomes visible, or it strengthens enough to support the weight of a (gentle) skier. So south-facing slopes, away from rocks + cliffs, and sunny days (and clear cold nights) tend to speed "resolution", while north-facing slopes, sheltered and shady, and cloudy days (and nights) tend to delay "resolution" of potential bridges.

There have been reports in cold winter of a deep unconsolidated snow layer collapsing under skiers down into a large crevasse.

(In late spring + summer the process and risk are different, but fewer people are out skiing then.)

  • Looking for signs of possible sagging bridges is usually not much help during normal skiing season. Because it tempts you to hope that a section not showing any sign of sagging is OK. But during normal ski season, bridged crevasses (whether safe or not) have any indentation filled in by seasonal snow, so there is often no visible sign on the surface of weakness underneath.

Of course if you do see a sign of sagging or weakness, then you should take it seriously. Non-skiing climbers are often advised to get good at looking for signs of sagging (and weak) snow bridges: Some of the hints are:  Difference in color: Brighter white because of more fresh snow accumulating in the slight depression. Or darker because more dust accumulated in the slight depression.

  • Multi-day tours: Need to be well prepared to handle a variety of situations -- because all it takes is a few hours with new snow and/or high winds to change everything.

  • [ serac-fall hazard has very different physics from crevasse hazard (even though the two sometimes occur on the same section of glacier). ]

  • [ avalanche risk in combination with crevasse risk . . . ]

A small surface sluff is much more dangerous around open or thinly-bridged crevasses, than  it would be on most other slopes. Even a very small or thin slide could be enough to take you down into a crevasse. Might not hurt to have a self-arrest tool ready.

Often glaciers have carved deeply into the surrounding rock, and there are steep slopes above the edge of the glacier, or even another glacier hanging above, waiting to drop deadly stuff down on you. Something to consider for route selection, and for where to not stop for a rest break, or even to take skins on or off.

approaches for getting evidence that crevasse are well-covered

(0) recent summer aerial photograph

This is a very helpful pre-requisite for any of the other approaches for getting evidence. Having accurate current knowledge of which sections of the glacier have wider crevasses, narrower crevasses, no crevasses, or which sections are "tormented" -- is very valuable. If you don't have it, you have to make guesses about the underlying state of different sections, in order to interpret the other evidences.

(a) sequence of aerial photographs for several weeks or months

If we had an aerial photo about once a week for say the previous 8 weeks, that could help a lot. Like if lots of crevasses in the section we wanted so ski were uncovered 8 weeks ago, but then 6 weeks ago there were all covered by a storm. At that point we don't know how strong the bridges were. If we also know that there were four significant storms since then, also some sunny days, and none of those crevasses were visible again, then we have evidence that the bridge has supported several layers of new snow, and we can hope that each new layer has had time to consolidate and added further strength to the bridge.

On the other hand if a number of the crevasses in the section we wanted to ski were uncovered until a week ago, then we have reason to be suspicious of the strength of some of those bridges. 

(b) trip reports about this glacier throughout this season

If we have a reliable trip report from 6 weeks ago saying that all the crevasses were covered, and another from 4 weeks ago, and another from 2 weeks ago confirming that they're still covered. If we also know that there were four significant storms since then, also some sunny days, then we have evidence that the bridge has supported several layers of new snow, and we can hope that each new layer has had time to consolidate and added further strength to the bridge.

On the other hand if we have a report from a week ago that a number of crevasses of similar size to what we might need to cross were still uncovered, then we have reason to be suspicious about skiing in that area now.

Note that a trip report from a single party saying only that they descended thru that section without any of the skiers punching through, is not such good evidence - (maybe they were just lucky). Unfortunately lots of trip reports are not very precise about what they did and what they saw.

Having several reports that many parties descended thru that section without any of the skiers punching through does count as useful evidence. But there could be a problem that you don't know if their routes were the same as yours -- especially if a new snowstorm covers all the old tracks.

(c) general seasonal info about snowpack + weather in this region

For popular touring routes, sometimes it can be known that the crevasses are well-bridged during certain months (e.g. late March thru mid-May for some sections on some tours) in a season of normal snow depth at the altitude of that glacier. If a particular season has snow-depth significantly above normal at glacier altitudes, that would be better evidence. It helps if the "knowledge" about this route has been confirmed in the last year or two by lots of ski touring parties crossing it without anyone punching through. Because glaciers keep changing.


How do you find out what "normal" is for that region at that altitude in that month?

Where do you get reliable info that no ski tourers punched through there in the most recent two or more "normal" years? (maybe a serious fall occurred, but the accident report was not put up on the web anywhere)

(d) at the scene: ask a local expert

It should be a real expert, someone who spends lots of days skiing and hiking and climbing on the same glaciers in summer and fall, not just winter - (not just another skier who's gotten lucky so far). A likely choice is a local mountain guide -- but they understandably want to get paid well for sharing their knowledge + experience.

Another possibility for glaciers near a ski-lift station is to ask the ski trail coordinator (or "piste coordinator" or "pisteur"). My understanding is that in most European regions the piste coordinator is not strictly responsible for crevasse risk off the marked ski trails ("pistes"), but I presume that they do have an interest in their station not getting a reputation for crevasse accidents.

(e) at the scene: ski descent tracks observed

Seeing lots and lots of skier descent tracks all over that section of the glacier does count as useful evidence that the crevasses are strongly bridged.

Seeing four skier descent tracks is not such useful evidence.

Seeing one skier uptrack is surely not sufficient evidence.

red light - yellow light - green light

red light

[ these are only proposed non-expert ideas, not reliable advice ]

  • December - January.

  • low snow year at high elevations.

  • You don't have current info about the underlying crevasses. You read about the route in a guidebook or web trip report. Other than what you can see on the surface of the snow in front of you, the best info you've got is an old 1:50000 map with a colored line drawn on it.

Lots of people do trips this way. Twenty years ago it was just normal practice for unguided parties. Nowadays it's not wrong to travel that way -- but if that's your approach, then you ought to have lots of other factors in your favor (see under "green light"), and very few other factors against you.

  • recent significant new snow with wind.

  • recent high winds - (which tend to blow snow around in ways that hide crevasses).

  • cloudy nights and days since recent new snow or high winds.

  • consistent cold temperatures since recent new snow or high winds.

  • daytime temperatures significantly warmer than yesterday.

  • first sunny day in a while.

  • glacier section known (from recent summer photos?) to be "tormented" (? large crevasses broken up in unexpected ways, seracs, etc. ?).

  • some crevasses nearby with width about the same as the width of many of the crevasses in the section to be skied (known somehow like from recent summer reports or photos) are currently not fully covered by snow.

  • reports or photos of this section just a few days ago said or showed that some crevasses were not covered.

temptation: People sometimes post trip reports with photos (or videos) of skiing fresh powder on a crevassed glacier the day after the snowstorm in February. But unless you know that section of the glacier is fully "plugged" (not real likely in February for some of the interesting high glaciers), it's usually a risky act -- even though somebody else "got away with it". And the higher risk often does not have sufficient reward, because fresh powder could have been skied on that day on some non-crevassed slope, and the spectacular scenery of the glacier could have been explored a later day after the fresh snow had consolidated.

red light behaviors - which could kill you even if lots of other factors are "green light":

  • taking air, or dropping onto the snow

  • taking off skis or snowboard - (even worse is walking with skis off)

  • grouping together on one spot when pause from skiing - (or during skiing)

yellow light

[ these are only proposed non-expert ideas, not reliable advice ]

  • late Feb - early March

  • late spring + summer

  • normal snow year at high elevation.

  • you have some knowledge of the underying crevasses from summertime photos or detailed diagrams or maps -- but the maps are ten years old, the photos are not shot from the most helpful angles or not really high resolution.

  • recent new snow.

  • only one good thaw-refreeze cycle (with clear cold night and warm sunny day) since recent new snow or high winds.

  • today's temperature and sunshine about like yesterday's.

  • glacier section known (from recent summer photos?) to have large crevasses (which a person could fall down a long ways before getting wedged, no possibility of "stemming" to climb out).

  • no visible skier tracks on this section since the last snowstorm, and no reports of skiing it in the previous three weeks.

green light

[ these are only proposed non-expert ideas, not reliable advice ]

  • April or early May

  • big snow year at high elevation

  • you have accurate knowledge of the underlying crevasses from recent high-resolution summertime photos shot with helpful angles and lighting (or possibly a detailed diagram, or recent 1:25000 map which shows crevasses in detail.)

  • several thaw-refreeze cycles since the last significant new snow or winds (like several clear nights and sunny days)

  • recent new snow came down warm + mushy - (so that in thaw-refreeze cycles it tends to soon collapse under its own weight or bond well to form stronger bridges).

  • glacier section known (from recent summer photos?) to have crevasses only of small size.

  • in springtime: daytime temperatures significantly cooler than previous day.

  • in springtime: night before was cold and clear.

  • in springtime: it's early the morning.

  • locations nearby known to have crevasses with width significantly larger than the width of all of the crevasses in the section to be skied (known somehow like from recent summer reports or photos) are currently fully covered by snow.

  • reports or photos of this section just a few weeks ago said or showed that all crevasses in this section were fully covered, and there's been three significant snowstorms since then with no significant melting.

  • this section is popular for skiers, and it's been skied many times in the previous month with no reported crevasse-related problems.

how crevasse risk is like avalanche risk

  • They both depend on structures hidden under the snow surface.

  • Wind is a key factor for the hazard.

  • Weather and snow history are key factors for the hazard.

  • Warm / Cold temperature cycles are key factors for the hazard.

  • There can be long periods of multiple weeks (or months) of elevated risk.

  • Early season tends to have elevated risk.

  • Late season tends to have more predictable risk.

  • Variability: Situations with higher risk tend to also have higher variability by location. (so in higher-risk situations, testing the underlying snow in one location does not reliably predict stability in a nearby location -- the terrain has dangerous "sweet spots" (or should they be called "bitter spots"?)

  • The greatest enduring danger after a snowstorm tends to be an "intermediate" situation:

With avalanche hazard the slopes that are fairly gentle tend to just consolidate without sliding at all, while slopes that are very steep tend to soon slough off naturally soon after the storm. So the higher risk tends to be on slopes which are steeper (e.g. 35-30 degrees), but not too steep.

With crevasse hazard, some snow types and terrain features tend to strengthen the bridge soon after the snowfall, while with other snow types and terrain features the bridge collapses soon after the snowfall. So the main risk is with bridges that are not weak enough to collapse, but not strong enough to withstand the weight (or impact force) of a skier, perhaps later (warmer) in the day.

crevasse risk is different from avalanche risk

  • crevasse-fall hazard tends to be more localized: often it's about the bridge strength over just one section of one crevasse. While with avalanche you worry about large slabs, or triggering from "hot rocks" far above. (Serac-fall risk is more about non-local physics). So I think the basic physics of crevasse-fall hazard tends to be a bit simpler than avalanche hazard.

  • careful use of procedures and equipment for roped travel and extrication can make the chance of serious injury or death from crevasse-fall pretty unlikely. While with avalanche hazard, if a big slab releases, or a slough over a cliff, you have a substantial probability of death even traveling with the best equipment and procedures.

see also

on the more on crevasse risk page:

and on other pages . . .


concept words: ski skiing snow skier skiers skis roberts report reports

backcountry mountaineering randonnee rando off-piste tour tours touring route routes

technique: techniques technical theory theories theoretical physics physical biomechanics biomechanical mechanics mechanical model models concept concepts idea ideas