PubTalk 3/2018 - Snow & Avalanche Science

Video Transcript
Download Video
Right-click and save to download

Detailed Description

Title: Snow and Avalanche Science - Highlights of applied avalanche research and forecasting

  • Avalanches impact transportation corridors, with subsequent economic ramifications, including the Going-to-the-Sun Road in Glacier National Park.
  • Large magnitude avalanches affect the landscape creating new habitat for flora and fauna.
  • Dendrochronology (study of tree-rings) is used to develop an avalanche chronology and examine contributory climate and weather factors.


Date Taken:

Location Taken: Menlo Park, CA, US


[rustling sounds]

[background conversations]


Welcome to the U.S. Geological
Survey in another installation –

or another installment – excuse me –
of our monthly public lecture series.

My name is Leslie Gordon, and it’s
always my pleasure to introduce –

not always – mostly my pleasure to
introduce the speaker for this evening.

So please welcome – but before
I do introduce the speaker,

I would like to entice you
to come back next month.

On April 26th, Curt Storlazzi,
who is one of our geologists

in our Santa Cruz office –
an oceanographer, geologist –

coastal and ocean – is going to
be speaking about the role of

U.S. coral reefs in coastal –
or, coastal protection.

How the presence of coral reefs
actually helps mitigate any

hazards of flooding and
wave inundation – storm waves.

So please do join us next
month to listen to Curt Storlazzi.

I have to slow down.


Thank you.

So tonight’s speaker
is Erich Peitzsch.

Erich’s a physical scientist in our –
the USGS Northern Rocky Mountain

Science Center in
West Glacier, Montana.

So I hope you’re enjoying our
sunny California weather.


At least it’s not snowy.

He’s been studying snow and
avalanches, and occasionally some ice,

with the USGS since 2007.
Erich began his snow and avalanche

career first as a professional ski patroller
alongside the great avalanche hunters

just up the road at Alpine Meadows
at Lake Tahoe, California.

Erich bounced around between research
and the operational world of avalanche

forecasting in Montana for a few years.
He earned his master’s degree in

on his Ph.D., both in snow science

at Montana State University
in Bozeman.

When not working, you might find Erich
chasing his two young energetic sons

around the mountains and
wondering where all his time goes.

So Erich will be speaking about
snow and avalanche science,

highlights of applied avalanche
research and forecasting.

So please join me in
welcoming Erich Peitzsch.


- Thanks.


Thanks, Leslie.
And thanks, everyone, for being

here tonight, taking time out of your
evening – this beautiful sunny evening.

They told me when I moved from
California to Montana that the quality

of the snow would make up
for the darkness. [laughter]

I’m on the fence still.

So again, thanks for being here.
And I also want to acknowledge

my colleagues in the Northern
Rocky Mountain Science Center,

particularly Dan Fagre, who is my
supervisor and leads our research group.

And then everyone there in our research
group. And as well as my colleagues

and co-authors on the work that
I’m going to present today – tonight.

So let’s do a little overview first.
And what is a snow avalanche?

Has anyone here ever actually
seen a snow avalanche before

in real – in real life?
[inaudible comments]

Sure. Pictures, videos.
All right. Well here’s another video.

And you can see the plume right there.
This in southwest Colorado

near Ophir, if you’re familiar.
And they – you can see the

avalanche fracture across the slope.
And it sort of looks like it’s really

slow-moving, but avalanches
can travel up to 100 miles an hour.

So what we’re looking at here is
the powder cloud of this avalanche.

And you can see it’s moving into that
forest off to our looker’s left side there.

Likely taking out
a few trees, maybe.

We don’t really know about this
avalanche, but we’re going to talk about

how avalanches can have an effect
on the landscape in this talk tonight.

And they delivered the –
or, they delivered explosives

via helicopter on this one, so no one
is actually in the starting zone.

And they do this because,
below this avalanche path is a

county road and utility infrastructure.
So they’re trying to protect the

county road as well as the
infrastructure at the bottom of this road.

I’m going to talk a bit about some
transportation corridors where

avalanches have an effect as well,
but in northwest Montana.

So, again, let’s talk a little bit
about the fundamentals here.

So we need four ingredients
for an avalanche to occur.

We need a sufficiently steep slope.
We need a weak layer, a slab

above that weak layer, and a trigger.
So in terms of slope, you know,

you can see on the snow on
this building that, at the top of

the building, it’s not steep enough.
It’s below 30 degrees.

Below the building,
it’s too steep.

You know, greater than 50 degrees.
But just in the middle is where

avalanches like to occur – pretty much
between 30 and 45 degrees.

When it’s too steep, avalanches
sort of just – the snow sloughs off.

And so we don’t necessarily
form a slab when it’s too steep.

If it’s not steep enough –
and we’ve seen avalanches

below 30 degrees on some slopes,
but if it’s not steep enough,

it’s just not enough for the slab
to actually move downhill.

And so what happens is,
we form a weak layer.

In this case, this is a layer called surface
hoar, and it forms on the surface.

It’s sort of the
winter equivalent of dew.

Via storm and transport of wind,
we have a slab on top,

and our skier comes along
and is able to trigger an avalanche.

And you can see that, when that skier
comes along and hits just the right spot,

perhaps it’s a convexity,
or a rollover on the slope,

where the stress is
actually concentrated.

The added addition of that stress
from that skier is too much

for that weak layer to hold.
So that weak layer collapses.

That fracture on that weak layer then
moves, or propagates, across the slope.

And so a trigger
can be a skier,

snowmobiler, snowshoer,
or it can be more snow.

So a weak layer can sustain
a slab to a certain point, perhaps.

And the weak layer either strengthens,
or we get an additional load.

So let’s say
it snows again.

Or let’s say we get a bunch of wind,
and the wind moves that snow

onto that slope and effectively
increases the stress on that weak layer.

At some point, our weak layer,
if it doesn’t strengthen, may not

be strong enough to hold up that slab,
and that’s when we get that collapse.

So why do we study avalanches?
Aside from, it’s a good excuse to

go out into the field and ski around?
We study because it is, indeed, a hazard.

Every year, the – or, the average
number of fatalities in the

United States each year is 27.
And, on an annual basis, that’s actually

more than earthquakes and any
other land slope failure combined.

Of course, earthquakes
are much more devastating,

as many of you know living here.
But every year, about 27 fatalities.

And it also has an effect on
transportation corridors.

So northwest Montana, on the
southern edge of Glacier National Park,

we have a canyon called
John F. Stevens Canyon,

named after the railroader
who put the railway in.

And you can see the
red arrow pointing to the railroad

just below these avalanche paths.
So if my cursor works, and it does,

we’ll – this is our avalanche path
right here – oops – right through

the middle between these trees,
and it comes right down here.

And you can see that they
actually built a snow shed

a number of years ago to
help protect against avalanches.

And we’ll talk about
why snow sheds can be effective,

and sometimes
they can’t be.

And you also see that the other arrow
is pointing to U.S. Highway 2.

So it’s a major transportation corridor –
shipping a variety of products

from the Midwest to
the coast and Seattle.

And we also have, again, the major
highway running through it as well.

And in the past, they’ve closed
the road due to avalanches

and avalanche activity.
And the alternatives are hundreds

of miles south or going into Canada
and going hundreds of miles north.

And so you can imagine
that it has economic ramifications

when they close
this section of road.

Even though it’s a relatively small
section, there are pretty big impacts.

So we also study avalanches because,
not only are they a hazard,

but they’re also a driver of
ecological and landscape change.

So this is a picture of bulldozers
and another machine moving avalanche

debris – not only snow, but a bunch
of big trees that were basically

destroyed and taken out by a
large-magnitude avalanche in 2009.

This is in Glacier National Park
along the Going-to-the-Sun Road.

So if anyone has been there,
likely you traveled it in the summer,

and there wasn’t a
whole lot of snow.

Certainly not avalanche debris in the
middle of the road – hopefully not.

And so we’ll talk about
some of the impacts that

avalanches have on
the landscape as well.

So first we’re going to dive
into avalanche forecasting.

So we’re with the USGS Northern
Rocky Mountain Science Center,

and we’re in a field station in West
Glacier, Montana, in northwest Montana.

We partner with the National Park
Service in Glacier National Park.

And we have an avalanche forecasting
program for the spring opening of the

Going-to-the-Sun Road in the spring.
So it’s almost time – around April –

the first week in April, April 1st,
where they start getting ready and

they start moving snow off the road.
And it’s – as I go to here –

it’s an 80-kilometer road, and

So the forecasting program,
we use a variety of weather stations –

automated weather stations.
We actually go into the field,

we dig pits in the snow to look at
the snow structure, and we help

forecast avalanches for the
road crew and other park personnel

to safely get
the road open.

So 11 kilometers of the road is
actually impacted by avalanches.

And there are 37 total avalanche paths,
as you can see on the map here.

Those green polygons indicate –
or, represent avalanche paths

that cross the road.
So on the image on the right,

you can actually see the road
as it crosses through right here.

It crosses through the middle of
some of these avalanche paths.

So in this case, the road isn’t
necessarily just at the bottom.

So we’re not just forecasting for
big avalanches that have, you know,

a probability of reaching the road.
We’re actually forecasting for

avalanche – even small avalanches
that might cross the road when the

road crosses through in the upper
part of the avalanche path.

So, again, it’s a vital artery of
visitation, particularly in the summer.

And that also, in and of itself,
has economic ramifications when it’s

not open for the local surrounding
community, but region-wide as well.

So the avalanche path starting zone
is along the Going-to-the-Sun Road.

The starting zones are where
the avalanche starts, right?

And that’s about 500 to 3,000
feet above the road level.

So you can imagine that,

often conditions can
be very different.

So it may feel like spring down
on the road at, say, 4,500 feet.

But at 7,500 feet, or 8,000, above the –
well above the road right up here in

this zone can be very different,
particularly the snowpack itself.

So, while down here, it might
resemble more of a spring snowpack,

up here it can still be a wintery
snowpack with quite a lot of variability.

So a little historical look at
some of the avalanches

along the Going-to-the-Sun Road.
In 1953, two machine operators

were actually killed. Three were 
caught, and two were killed,

in a wet snow avalanche that released
in an area called Alps 1 avalanche path.

And you can see the debris here and
folks actually trying to dig through.

And that was a pretty
destructive avalanche, of course.

In 1964, a bulldozer was knocked off
the road in an area called the Big Drift.

And this is pretty much right at the –
at Logan Pass, which is

right on the continental divide.
So this is actually just on the east side.

But you can see over here,
this is the crown of the avalanche

where we’re looking at about 12 to

So, as this bulldozer was moving
through, it actually triggered the

avalanche itself and then
was knocked off the road.

And the most recent was in 2005,
which was a close call.

Fortunately, no one was injured, but a
similar situation where the bulldozer was

moving snow, pushing snow off
the road, triggered an avalanche.

It broke right next to,
or right at the bulldozer,

and the bulldozer
went over the road.

Fortunately it was a
very skilled operator.

He was able to put the blade down,
and it didn’t roll the machine or

didn’t go that far, and they were able
to pull it back up in a couple days.

So, as I mentioned earlier, we have
the partnership with the Park Service.

And we, as USGS researchers,
we try to figure out, how can we

make forecasting better so
that we’re keeping people safe?

And the biggest problem that the
road crew faces and that

people traveling along the road
in the spring face is wet snow.

Right? We’re moving into the spring,
so we’re not dealing so much

with dry snow avalanches,
though we do when it does storm.

We’re really focused on the
wet snow avalanche research.

So let’s talk a little bit about
what wet snow actually is.

And there’s three main types
of wet snow avalanches.

We have wet, loose avalanches,
wet slab, and glide avalanches.

So wet, loose avalanche –
as you can see from the image here,

these avalanches involve
really just the surface snow.

So anywhere from the top few
inches to maybe a foot at most.

But what happens is, they start
at a point, and then they tend to fan out,

as you can see, like in
an upside-down V pattern.

And, in this case, sun is shining
on these – this is a southerly –

or southwest-facing aspect,
so we have a lot of sun hitting this slope.

It’s warming up the rocks.
And you can see that all of

these avalanches in this
image started near these rocks.

So we have a lot of short-wave
radiation being absorbed by the rocks,

heating up the snow adjacent to it,
and that’s where we’re going to

get these wet, loose avalanches.
So what happens is, as the sun and

warm temperatures heat up the snow
on the surface, it reduces the strength,

or reduces the bonds,
in between each little snow grain.

And then it reduces its strength,
and it releases and goes downhill.

Because that’s what
snow ultimately wants to do.

And before I go to this one,
the wet, loose avalanche – you know,

they’re relatively predictable.
We can say that it’s going to be

warm and sunny, and we’re probably
going to see wet, loose avalanches.

But slab avalanche, like in this image,
is definitely harder to predict.

And this is a really large avalanche.
As you can see, it’s sort of – this is the

crown right over here, starts up here and
comes all the way down and around.

And what happens here is,
as I mentioned before in a

dry slab avalanche, we have
avalanches occurring because the

weak layer can’t hold the slab above.
Well, that’s the same in a wet slab

avalanche, but instead of adding
more stress or a load to the slab,

we’re actually decreasing the strength of
the weak layer in a wet slab scenario.

So in the schematic here on the right,
as water moves through the snowpack,

whether it’s melting or whether it’s
rain on snow, it moves through

the snowpack via flow fingers.
It’ll – or, water will move through

the snow pack any way it can –
the least-resistant path.

And it’ll eventually hit, potentially,
an interface between the slab and,

if we have a weak layer,
in the snowpack.

And with fine grains
over coarse grains,

what it tends to do is
pool along that interface.

When it does that, it decreases
the strength of that weak layer,

and it’s not able to
support the slab above.

So that’s a
wet slab avalanche.

And a glide avalanche – we sort of
call these the circus oddity

of the avalanche world because
they’re really hard to forecast.

So as I mentioned, snow likes to move
downhill. Gravity likes to do its thing.

And it moves downhill at variable rates.
And we call that glide.

So, as the snowpack moves
downhill at variable rates,

we tend to get a
tensile crack that forms.

And so here we have what we call
a glide crack, or a tensile crack.

And you can see, sort of in this area,
and all the way around,

the snow actually buckling.
And what can happen here is,

when we get a glide crack, often we’ll
get a glide avalanche, but not always.

But what we need is free water
at that ground-snow interface.

So as water moves through the
snowpack, it then pools at the bottom of

the snowpack and the ground interface.
And it basically lubricates that interface

right there, so the snowpack will then
start to glide downhill even more.

And it also requires relatively
minimal surface roughness.

So if we look at this image,
and you can see these are glide cracks –

one’s right here, over here,
another one here, up here.

You can see that they’re all in rock slabs.
So you can imagine that, as the

water moves through that – or, that slab,
when it hits the rock slab and the

snow interface, it’s not like soil
where it can actually move into the soil.

It actually hits the rock.
It’s got nowhere to go.

So it’s going to move
along that interface.

So we tend to see glide avalanches
occurring on smooth rock slabs.

And in the same
location every year,

which we’ll get into in
just a little bit here.

So we wanted to start at the beginning.
You know, what’s actually causing

these wet snow
avalanches to occur?

So we looked at wet slab and glide
avalanches because we – again, we have

a pretty good handle on what
happens with wet, loose avalanches.

And so we looked at –
on the right is a – the results of a –

what we call
a classification tree.

And we looked at avalanche
days versus non-avalanche days.

And we looked at the 60 variables
of weather and snowpack to

basically determine, what are
the most important variables?

Or what are the discriminatory
variables between an avalanche day

and a non-avalanche day?
And so what we found out is the

first node here – avalanche days
and non-avalanche days will split,

and avalanche days we get
with maximum air temperature.

And, as we move down, again,
non-avalanche and avalanche days,

the non-avalanche days that were left
are split again on mean air temperature.

And then the final node right here
splits on a change in snow depth

over a period of five days.
So what this tells us is that,

one, air temperature is important.
And two, the settlement in the

snowpack seems to be
an important variable as well.

And so this is really important because,
as we see increasing air temperatures,

particularly, you know, creeping into
late winter, we may see – we may

begin to see more wet snow avalanches,
so it becomes even more important

to try and understand
these wet snow phenomena.

So, as I mentioned, these glide
avalanches tend to occur in the same

location every year in areas that affect
the road – transportation corridors.

We call these repeat offenders.
And these repeat offenders, we wanted

to figure out, well, why are they
occurring in the same spot each time?

So we looked at, again, a number of
variables, but this time terrain variables.

So slope, curvature of the slope,
substrate underneath – we were looking

at smooth rock slabs versus, say, areas
with a lot of vegetation underneath.

And what we were able to find is that,
of course, as we – as I just pointed out,

the areas with smooth rock slab
are areas where we tend to see

a lot of glide avalanches. So we modeled
this across Glacier National Park.

The red polygons indicate areas of
known glide avalanche activity.

And the blue indicate modeled areas.
And so we can see that there’s a

fair amount of area that has the
potential to harbor glide avalanches.

And we also use time-lapse
photography to get a handle on

when these things are actually occurring.
Because we’re not out there 24 hours

a day, and of course we can’t capture
it at night because the – it’s dark.

But we can see here, just as an example,
where the red box highlights,

you can see, two days later, we have
what appears to be a small avalanche.

It’s actually about 200 to 300 feet
in width, but we’re pretty far away –

or, the camera is
relatively far away.

It’s actually about

And then, even two days later, we have
another small glide avalanche right here.

And you can see the glide
crack begin to form right back here.

And then, in just one hour, we have
a really massive glide avalanche

that occurred in
this basin right here.

So it’s – again, time-lapse photography
provides us a nice way to be able to

capture these events when we’re not out
there and we can’t see them in person.

So let’s shift a little bit from
avalanche forecasting, and we’re

going to talk more about tree rings
and avalanche ecology and how we

can use tree rings to actually
develop an avalanche chronology.

So dendrochronology,
if anyone here or online is familiar,

it’s the study of tree rings.
And so we can actually use tree rings –

you know, people use tree rings to
sort of study how – perhaps how big

of a snow year it was, right?
We get big tree rings sometimes.

Or if we’re in a drought, you know,
we’ll have often really thin tree rings.

So we can use these tree rings and
count them by actually looking at –

avalanches, when they hit a tree,
will leave a signal.

And I’ll get into that signal in just a
second, but first, let’s talk a little bit

about the avalanche path morphology
because this is important.

As I mentioned before, the starting zone
is where avalanches tend to occur.

Avalanche will release in the starting
zone and move down this track.

And then it’ll eventually
deposit debris in the runout,

or the deposition, zone all
the way down here at the bottom.

And so this is an image of the avalanche
path that I showed earlier where the

bulldozers were moving a lot of debris.
This is in Glacier Park.

And this was – this is the image –
this is an actual image taken

after that large-magnitude event.
So all this brown color you see here

is actually vegetation and debris –
trees knocked over and transported

in this large-magnitude event.
A lot of these trees were actually

carried from way up here.
So right here by this little nick point,

this was actually forested.
And it increased the avalanche path.

This large-magnitude event increased the
avalanche path by about 30% in width.

So you can see here that this track is,
in this particular avalanche path,

is relatively low-angled.
It’s about 15 to 20 degrees.

And most avalanches, if they’re small, or
maybe even medium, in size will begin in

the starting zone and probably end up –
and typically end up right around here.

So from this starting zone,
to give you a sense of scale,

all the way to the runout zone,
is about 4,000 vertical feet.

And so, to begin – an avalanche
beginning way up here, to cross a track

that’s relatively low-angled,
and then to take out a lot of trees

and end up all the way

or the valley bottom,
is a really big avalanche.

And so we’ll talk a little bit
more about that one in just a sec.

But getting back to tree rings.

They’re kind of the fingerprint
of avalanche activity.

You can see here that this sort of shape
right here and this shape of the tree,

this would be the uphill side of the
tree before it was knocked over.

So when an avalanche comes down,
hits a tree, it’ll impact that tree.

And it’ll actually – not always,
but often leave a scar on the tree ring.

So here – and this blue arrow
is pointing to a scar from 1993.

This one is a little scar but more
of what we call reaction wood.

And this one over here is, again,
a little bit more of reaction wood.

And so there is a signal that it leaves,
and sometimes it’s a

really good signal, like a big scar.
And oftentimes, it’s a little more subtle.

So for this project, we actually
are in Glacier National Park,

but also in the surrounding
adjacent mountain ranges.

So we’re looking at
four mountain ranges here.

The red dots
indicate our sample sites.

So we collected over 600 samples –
these cross-sections from trees

so that we can actually look at the –
look at the tree rings.

And a quick little overview –
two of our sites, again,

are in John F. Stevens Canyon.
The blue polygons are the avalanche

paths, and then, again, the map that I
showed earlier, just in a different color,

of the avalanche paths that affect
the Going-to-the-Sun Road.

So here’s a little video about
what we actually do in the field.

Oop. And let me – let me see
if I can plug in [chuckles] –

I forgot to plug in the audio here.
But we’ll continue on.

Basically, what’s happening –
and I’ll let it run –

is we’re looking for
scars on the tree.

And so, in this case right here,
we suspect that that’s a scar –

or, that hit the tree, and what
we’ll then do is cut into the tree.

We want to get a cross-section
from this tree, and so we take out

this cross-section.
We could core these trees, but they’re

dead and down, so it’s okay to take
a cross-section from these trees.

But often what happens,
if we just get a core from this tree,

we’re not able to –
we might miss the signal.

And so, you know, depending on
where we core – even if we core

in all four, or more, directions,
one, that’s a lot of cores.

But some of these scars are really small.
And so we might actually miss it.

Even if we go in where the scar is,
on the outside – the exterior of the tree.

But again, not all trees are
going to show a scar on the exterior.

So maybe an avalanche –
a really big avalanche hit in, say, 1950.

And for instance,
let’s look here.

We have avalanche scars from ’72, ’85,
and ’93 on this 83- or 82-year-old tree.

So an avalanche came in
and hit this side.

And then it looks like it
also came in and hit this side.

So we sort of have the tree rings
growing around that scar –

around that disturbance.

But imagine that maybe this wasn’t here,

and we didn’t know where the scar
was on the exterior, and we might

have missed – by coring,
we might have missed

a smaller scar that exists further in.
So having these cross-sections,

while it is a lot of work,
because we actually have to

sand them down and
make them really smooth.

Because often you can’t see
these scars until that tree –

that cross-section is quite smooth,
and then they pop out.

And so for instance, here’s one where,
in 2003, we had an impact scar.

And, as I mentioned before, there are a
variety of – or, there’s a degree of signal.

And in 1959, we have something –
what we call reaction wood.

So we might not necessarily
have a really obvious scar.

But when it gets hit by a tree on
the uphill side, the tree – it counteracts

that and buffers that by growing
more wood on the downhill side.

They call that compression wood.
Or, if the rings are really small,

we call that reaction wood.
So we don’t necessarily get a scar.

We have other things
to sort of fall back on.

So not only are we able to build
a chronology over time looking at

sort of the large – or, the magnitude of
avalanches over time, but we’re able to –

when we collect trees in place,
we’re able to develop a

return period frequency map.
So this is near the railway in the –

or, this is the railway in the
U.S. Highway 2 corridor.

This black rectangle here is a snow shed.
And what we’re able to do then –

again, with our in-place samples,
is we’re actually able to look and

develop a return period frequency.
So you can see down here, obviously

avalanches will reach the railway
grade or the road much less frequently

as they do further up in the path.
You know, up here, it’s anywhere on

a scale from about 2-1/2 to 4 years.
Whereas, closer to the snow shed,

you know, we’re looking at about
a 6- to 10-year return frequency there.

So these are just three avalanche paths
along the Going-to-the-Sun Road,

which gives you a bit of a sense of, you
know, how old our avalanche paths are.

The problem with studying trees
in an avalanche path is the avalanche

likes to take out the trees.
And they’re not always there, sitting

there nice at the bottom of the avalanche
path, or in place, for us to just sample.

And so we have, you know,
regeneration of these –

of these forests,
and a lot of these trees are young.

So sometimes it’s hard to get a
chronology that goes back very far.

Fortunately, so far,
we have a decent chronology.

And, in this case, you can see,
this is the number of responses, actually.

So this isn’t the
number of samples.

We’re looking at avalanche
responses in these samples.

And we actually have responses
that go back to the late 1700s, which,

in an avalanche chronology,
is pretty far back.

You know, folks who study
other proxies may not think so,

but for an avalanche path,
that’s actually pretty far back.

So what we’ve done here –
as you can see,

there are more
responses in recent years.

And that makes sense.
There are more trees.

So in order to normalize it or to
scale for it, we actually take the

number of responses per the number
of trees actually alive in that year.

And so we then have a
threshold percentage that

we use to say, well, this is,
you know, likely an avalanche here.

So aside from developing a chronology
and a return period frequency,

avalanches also have
an impact on the landscape.

So you can see from these repeat photos
in 1997 to 2004, the number of trees,

of course, is markedly different.
But one thing to notice over here

is that there are – there are
a fair number of trees, but then,

over on this image
in 2004, there aren’t.

And you can see that in a large event,
we could potentially have an avalanche

that comes down, and if it’s the whole
width from what we call the trim line

over here to the trim line over here –
if it’s the whole width of this avalanche

path, it’s going to come down
and potentially reach the rail grade

on the side of
the snow shed.

So, as I mentioned earlier,
snow sheds can be very effective.

But when we have a changing
landscape like this, they’re

not necessarily 100% effective
from guarding against avalanches.

So, in this case, we also have
a interaction with wildfires

where wildfires – you know,
if they burn the forest,

they open up, potentially,
more avalanche terrain.

Trees act as not only
an anchor in the snowpack,

but they can also act
as surface roughness.

When an avalanche comes down the
slope, and it hits a bunch of trees, it can

actually – you know, the trees help slow
the velocity of that – of the avalanche.

But when there aren’t any,
because they’ve been taken out

by an already large avalanche,
it makes the probability of an avalanche

reaching further down even greater.
So you can see that the difference

in landscape change and how that
might affect the avalanche regime.

And, again, you know, looking at
return period frequencies, this is just

another illustration where,
when we – these are all

downed trees from an avalanche.
So when we take out these downed trees

from a large avalanche, perhaps
what was maybe a 2-1/2- to 5-year

return period frequency up here
is now perhaps down here.

And perhaps the 10-year now
becomes even further down.

So by developing an avalanche
chronology, one of our goals in this

study – and we’re in the process of this
work right now – is to actually associate

some of these large-magnitude event
years with weather and climate drivers.

So looking at things like, you know,
storminess or, you know, upper-level

air pressure indices. And potentially
other teleconnection indices as well.

And what’s sort of noteworthy
about this cycle – and this was the cycle

in 2009 that I previously mentioned –
is, in early January, we had,

you know, relatively cold
temperatures – minus 15 degrees C.

And we had a pretty sharp increase
in air temperature to near-freezing.

And this is at about 7,500 feet.

is pretty much the alpine.
So it’s, you know, right at the

upper level of sub-alpine and
right getting into the alpine terrain.

And so that, for us,
is a pretty high snow level.

You know, it’s pretty similar
to here in the Sierra as well.

Where if you get
snow levels at, like, 8,000 feet,

that would be
fairly comparable.

Or at least in the Tahoe region in the
high Sierra, it’s a bit different, of course.

But you can also see the –
during the cycle, we had

a sharp increase in precipitation as well.
And so it rained at the low and

mid elevations, and even
a little bit into the upper elevations.

So what we had is a lot of rain on
snow and a potential wet slab scenario.

But then we also had really
heavy wet snow up high,

sort of acting as a dry slab scenario.
So we have sort of the combination

of both wet slab and
dry slab in this same avalanche.

And the image on the left is –
it’s actually from a different avalanche,

but it’s from a
wet slab avalanche.

And you can see that the debris that
it can deposit is pretty substantial.

So in that 2009 avalanche,
the image on the left shows the

debris pile left along the lower part
of the Going-to-the-Sun Road.

Again, about 4,000 feet below.
And the image is foreshortened there,

but that debris pile is about 30 feet deep.
And that person is definitely not 10 feet,

so it’s a little foreshortened.
But then, on the upper part of the road,

it actually damaged
part of the road.

So it took them a while
to repair that that summer.

All right. So another project we’re
working on is snow depth mapping.

And we want to know the,
not only variability of snow depth

across an avalanche path starting zone,
but we also want to get a sense of,

you know, how much
it’s changing over time.

So when we get a storm, you know,
we have automated weather stations

that sort of give us an
idea of how much it snowed.

But then, when we have wind events
that tend to load snow from perhaps

the leeward or the windward –
or, the windward side of the ridge,

loading the leeward side, some –
you know, what we do is basically

make a good educated guess about how
much new snow has been transported.

So we might say, well, we had
winds for, you know, six hours

in the 30-mile-per-hour range,
and now we have a slab that’s

about maybe a foot in addition
to what we already had.

So what we’re trying to do is
get a better sense of how much

things are actually changing.
So we’re using drones, or unmanned

aerial systems, and a photogrammetry
technique called structure from motion.

And we’re basically flying over these
starting zones and mapping them.

And I’ll explain
how we can do that.

But first, structure from motion
is a photogrammetry technique

where you take our – in the middle here,
our scene that we want to –

we want to
image and map.

And in our case, you know, we don’t
need to map the back of the avalanche

path or the back of the mountain,
so we just do about a 180 around it.

And what that allows us
to do is get overlapping images

of our scene of interest.

And the objective is to detect change
in snow depth through time so that

we can create these high-resolution
products where we can actually

look at a change in snow depth from,
say, one week to another week.

Or even just a few days if we
have a storm or a big wind event.

So we take – we take our drone out,
and we just have a small quadcopter.

And we’re able to attach our
camera to the bottom of this.

And it’s a 15- to 20-megapixel camera,
depending on which one we’re using.

And we’re actually able –
in this case, we were mapping

the open area to the –
to the right over there.

And while it doesn’t look like
your classic avalanche path,

it is avalanche terrain.
It’s sufficiently steep,

but it’s a little more forested.
So we wanted to get a sense –

can we – you know, we have a
pretty good sense that we can capture it

in this open terrain, the snow depth
change, but we wanted to look at,

are we able to capture it in
more forested or sub-alpine terrain.

So here you can see, in the bottom right
where those two green markers are is

where we launch and land. And the
other lines indicate our flight paths.

And so this is just a small screen shot
of when we plan our missions in the

office before we go out and actually fly.
We want to be sure that, one, we’re,

you know, well above the –
above the ground.

So we use a digital elevation model,
but we also have to account for

trees that are
out there as well.

So we – you know, we have to –
we have to plan it pretty carefully.

And we take, you know,
a cross-section – or, a upslope

and downslope and then
a cross-slope transect as well.

And we’re limited by battery
power on this particular drone.

And, you know, it’s a quadcopter,
so our battery life is about 12 minutes.

So we’ll typically fly in upslope and
downslope transects, bring it back,

and then send it back out again
after we swap our batteries.

So if we go out and we
just try and map this, that’s great.

We have an idea of what think –
you know, or what shows how much

snow depth is and the variability.
But we actually need to go out and

make sure and verify that
that’s the case, particularly now.

So we go out and we do snow
depth manual measurements.

Again, we don’t want to necessarily
put ourselves in the starting zones here.

So what we tend to do is walk around
on the safe ridges around our avalanche

path, and that’s still within our
scene that we’re interested in.

And so we’re able to verify
these snow depths by probing

and taking
manual measurements.

We also have to place ground control
points out there so that, when we take

our images, that they need to be visible
in a number of our photographs.

And so we put these – we spray
paint these X’s on the snow.

And we take manual measurements
there too, but we get high-resolution

GPS and GNSS points so that we’re
able to actually georeference our images

and our maps, of course, once we
bring it back into the office and

do all of our processing
so that it’s even more accurate.

So going out in the field is really fun.
You know, that’s the fun part.

That’s why we do the work.

But the processing, of course,
is what takes the longest.

And the nice thing, though, is that we
put all the time in by processing and,

you know, let our – let our –
let our computers do a bunch of

the work and
let it run overnight.

And we’re able to create these really
cool and high-resolution products.

This is a dense point cloud.
So if anybody is familiar with Lidar,

you know, you fly over an area,
and Lidar basically shoots a –

the sensor shoots a laser down,
and it gives you a return pulse,

and you can basically –
it gives you a bunch of points.

And you can then get a bare earth
digital elevation model with Lidar.

Structure from motion
works similarly.

As I mentioned, we take
these overlapping images.

And so what it does is, it takes
the common or top points in

those overlapping images,
and you can imagine that that sort of

propagates through every image as
you have these common points.

So, in this dense point cloud,
we have about 50 million points.

And it’s a relatively small slope,
even though it is, you know,

an avalanche terrain.
The vertical relief here is

about 500 feet,
so it’s not quite that large,

but 50 million points is, you know,
when processing is – can be fairly large.

And so the difference with structure
from motion is that, when Lidar

moves through, it can actually –
moves through gaps, you know,

in trees, and it’ll create
a bare earth model for you.

Structure from motion isn’t
able to actually move through.

You know, it’s not
sending a pulse down.

It’s using these tie points.
So we can’t create a bare earth

elevation model, but we can
create a digital surface model.

And so here we’ve created a 3D model.
And it’s really low-resolution.

The high-resolution one
that we created would have

crashed the presentation,
so I didn’t want to put it in here.

But 3D models are nice.
You know, they’re pretty to look at.

They’re good for outreach
and presentations, but really,

it’s the digital surface
model that we’re interested in.

And you can see – this is a
digital surface model

with an accuracy of
about 4.9 centimeters.

And so it’s good enough
that you can actually see –

right here, you can
see our ski tracks.

And we made nice, pretty figure 8s
coming down. [laughter]

And over here, you can see
sort of our up-track

where we skimmed up to
take our manual measurements.

And down here on the far right,
you can actually see our vehicle.

So this is important because,
with a 5-centimeter accuracy

in both the X and Y and even
the vertical realm, we can actually

detect change on a really small scale.
So, yeah, we’re going to be able to

detect change when it snows 2 to 3 feet,
but we can hopefully also detect

this change when we have,
you know, storms or wind events

that are around a foot,
or maybe even a little less than that.

So what we then do is take our digital
surface models or our point clouds

that I showed earlier, and we basically
take the difference between the two.

It’s simple subtraction of those points
in the point cloud or the pixels

in your digital surface model.
And that allows us to get our change.

And, again, this is a great tool.
Because now we’re able to, you know,

launch from all the way down here,
not necessarily put ourselves in a

hazardous position, but we’re able to
map a really big area and look at the

variability across that area as well.
And that helps with, you know,

forecasting for, okay,
we got a wind event,

so how much more of a slab
did we put on top of

that weak layer, if there is one?
Or how much new snow did we get?

Another nice product that we’re able –
or, another byproduct is we’re able to

map these
avalanche crowns.

And so this is a digital surface
model that was created last spring.

And you can see here this arrow points
to an avalanche crown right here.

And the accuracy of this digital surface
model is about 13 centimeters.

And it captured a small one
over here as well.

So in real life, this is what it looks like.
This is the avalanche crown that we

were able to capture, and then
this is the small one over there.

So any guesses as to the height of the
crown of this avalanche right here?

No guesses?

- Thirty feet.
- [inaudible] the trees. It’s [inaudible].

- Thirty feet?
- Little higher.

- That was a guess. Higher?
- Higher. Much higher.

- Much higher?

In the last talk, anyone – we got
a range from 2 meters to 200 meters.

So it’s somewhere in between there.

So, yeah, around here,
it’s about 15 to 20 feet.

And then over here, at its farthest
edge where this crown is, it’s upwards

of 60 feet. So that’s a big –
you know, that’s the whole snowpack.

We’re looking at a
glide avalanche right here.

And, you know, this is an incredibly
deep snowpack because what’s

happening is, over the course of
the winter – and this is in the spring,

but over the course of the winter,
we have, of course, storms falling,

but then we have wind
loading from sort of the

back of this image,
as wind transports snow.

And it just basically deposits
all that snow right down here.

And we also have
avalanches that occur up here

and put all their
debris down here as well.

And so it’s an incredibly
deep snowpack over here.

But this tool allows us –
because this is quite far away,

this tool allows us to actually capture
how large these avalanche crowns are.

And when it goes to the ground,
you know, that’s obvious.

But when we’re working with crowns
that aren’t necessarily at the ground,

it gives us an idea of what
layer they actually released on.

All right. So to finish up here, you know,
our next steps is our research is going to

continue to be driven by operational or
decision – management decision-making.

So we’re there to basically help –
you know, in this case,

on the transportation corridors,
we’re there to help keep people safe.

We’re going to continue
to refine our wet snow models.

So, you know, we used a
number of variables as inputs.

But we also want to continue down
that road and make sure that, you know,

the variables we put in and what’s
coming out still works and it –

and we’re using that operationally
to some extent as guidance as well.

And then, this spring, as we –
if it ever melts in Montana –

it’s been a big year – a big snow year.
But when it does melt, we’re going to

continue to explore and map our
snow depths across these starting zones,

not only in periods of
accumulation, but also melt.

So look at how rapidly
the snowpack is settling.

Because, as we learned before,
that was one of our important

variables for our
wet slab avalanches.

So thanks for attending.

And any questions?


- Thank you, Erich. And I’m sure
many of you have questions.

And many of you have been here
every time, so you know the routine.

You’re first in line. We have two
microphones set up in the two aisles.

Please, if you are able, get up and use the
microphone so that we can all hear you.

If, for some reason,
you’re not able to get up

to the microphone, wave at me,
and I’ll come and bring you one.

So why don’t you go ahead, sir,
with the first question?

- Thank you for the presentation.
I had a question about pre-empting

these avalanches through blasts.
- Yeah.

- The way they do on ski slopes.
So if you knew that, say,

the BNSF railroad was in the –
in the path, could you potentially

pre-empt that or trigger
that avalanche before?

And if so – if not, why not?
- Yeah.

- Thank you.
- That’s a great question,

and I should have
elaborated on that one.

But – so, yes, obviously,
you know, using explosives

is a great mitigation tool.
So there’s a couple of things there,

is, one, yeah, you can close the road
and the railway for a short period of time

and, you know, deliver the
explosives in a variety of techniques.

And, again, you know,
create the avalanche on your terms,

when you want it
to go down.

One of the issues –
and they do that along the railway

in emergency purposes – so if it’s a really
big storm, and avalanches are imminent,

and they’ve already seen
smaller avalanche activity.

But one of the issues is that the starting
zones are in Glacier National Park,

which is a wilderness area.
And so they did an entire

environmental impact – or, EA –
environmental assessment, and they –

it was deemed that they weren’t
going to allow avalanche mitigation

via explosives on a regular basis
and only in emergency conditions.

So that’s the reason now that the –
that they don’t do it regularly.

Because you’re right.
You know, if you – if you –

basically, at every storm,
even if it’s a small storm,

you’d basically mitigate it through
explosive mitigation then, you know,

you could – you could knock down
smaller avalanches and potentially

have less – you know,
less of an effect.

But because the starting zones are in
the park, that’s what they decided.

- Hi. I’d be interested to hear anything
more you might say about how global

warming is affecting your work.
- Yeah.

So that’s a – that’s a great question
that I actually get quite often.

And it’s one of the questions
that we’re working on.

So obviously, you know, global
air temperatures are increasing.

And as I – as I sort of alluded to,
we might – we might see more

wet snow avalanches
creep into late spring.

And perhaps even more
rain-on-snow events mid-winter.

You know,
especially here in the Sierra.

I mean, you know,
we’re used to rain-on-snow events.

And in the Cascades –
but those coastal regions.

But now, you know, we may begin
to see – and I say “may” because,

at least in the U.S., there hasn’t
been any definitive work.

So that’s part of what we want to try
and look at with the tree ring work is to

see – we can’t necessarily tease out, you
know, rain-on-snow events mid-winter.

But we can sort of determine, at least on
a shorter scale, where we have weather

records that – you know, automated
weather stations that sort of show that.

Just north in British Columbia, north of
us, near Rogers Pass they looked at,

you know, some of this,
and they found that there is

a slight increase in
mid-winter rain-on-snow events.

And their observational record goes back
a lot further to the, like, 1940s or ’50s.

And then, in the Alps,
you know, they’ve shown that

there has been no change.
But, again, you know, observational

records can be pretty scarce and hard to
come by as we go further back in time.

So that’s why we’re trying to use
tree rings as a – as a proxy there.

- I had a question.
So I was wondering if drones

would have any role in delivering
explosives for – and how does

your work compare with
what’s going on in Europe?

They’ve built things in the path of –
we’ve got a head start as we haven’t

built in front of these things as much.
- Yeah, exactly.

- But I was wondering how this work
compares to what’s going on over there.

- Yeah. So the first question of
delivering explosives via drones,

I actually read something a few
months ago about a company that was –

that proposed to do that,
or they were trying to do it,

outside of Telluride in Colorado.
I haven’t heard anything since.

And it’s doable.
As long as the payload, of course,

on the – on the drone is –
or as long as the drone has a,

you know, pretty high payload, and
depending on what sort of explosive.

You know, if it’s a 2-pound charge
that a lot of skiers use – you know,

just a stick of explosive,
then that’s certainly doable.

But other than that, I haven’t heard.
And, yeah, it might be – and in terms of

the work that we’re doing, you know,
comparable to Europe, in that – I mean,

they have – you know, yeah, they’ve
been living and working and dealing

with avalanches and avalanche terrain
all the time – or, for a long time.

And they have a variety of techniques
that they use to mitigate avalanches,

from explosives to – you know,
they basically will build structures

in the starting zones to disrupt the
snowpack so that that weak layer

can’t form – you know, or can’t
form widespread across the slope.

So they’ll use snow fencing.
They also have avalanche dams.

Like, for instance, in Innsbruck
in Austria, they have avalanche paths

that come all the way down into town.
But they have these really big diversion

dams sort of higher up, of course,
so that, when an avalanche

comes down, it’s diverted.
And so you don’t have the

concentration of impact,
and it sort of disburses that impact.

So they – and, you know – and in
Colorado too, there’s a lot of work

being done there in terms of, you know,
building structures that help mitigate

and also explosive delivery –
remote explosive delivery too.

Not just throwing charges.
- That was kind of the basis of the

question is that you could get access – 
well, quickly and to areas that people

wouldn’t want to – well, you just
wouldn’t get into normally.

- Yeah.
- Or wild areas.

- Yeah, exactly.

- It seems to me this is similar
to fracture in brittle materials.

- Yeah.
- And I wonder if you’re using

the methodology of fracture mechanics
to look at stress intensity and critical

flaw size and all that sort of thing?
- Yeah. Yes. [laughter]

- Yeah.

- I’ll explain more than that.
So I – let me caveat that with,

I am not an engineer.
But, yeah, a lot of – a lot of folks,

particularly, again, in Switzerland and
folks out of Bozeman have looked at –

have used fracture mechanics to look
at fracture within the weak layer.

So, you know, one of the things – and it
was actually a really cool test we use –

when we dig snow pits, we do
something called a stability test.

So you can basically isolate a column.
And previously, we would just isolate a

So 1 foot by 1 foot.

And we’d take our shovel – our
avalanche shovel and set it on the top.

And you sort of tap on it
to apply stress, right?

And so you’re looking to
see if that weak layer, or any layer,

in the snowpack breaks.
So about 10 or 12 years ago,

sort of using mixed-mode propagation, 
where we’re not just looking at –

we’re not looking at shear.
We’re also looking at collapse.

And so what they found is, if you
take a block, now 30 centimeters back,

but 90 centimeters wide, we can actually
look at not just the fracture part of

that equation, but also propagation.
And that’s really what we want to

know is, yeah, it might break, but is it
actually going to propagate across?

So, again, we’re using a small block
that may or may not be representative.

It’s just a small point.
Especially with spatial variability.

But it gives us a better idea
of how likely is that weak layer

to not just fracture
but also propagate?

So, yeah, those guys have been doing a
lot of work with fracture mechanics.

- Thank you.

I was curious why you
chose the photography over Lidar?

- Yeah. So the – right now –
we actually have Lidar – or have

flown Lidar along the entire canyon.
That was in the summer.

So we have a reference –
digital model that we can use.

Problem with – we want to – we want to
sample on a relatively short time scale.

So we’re talking, you know, maybe
every week or every couple weeks,

or even more frequent than that.
To be able to – you know, typically,

Lidar is flown from
a fixed-wing aircraft.

And so, as it’s flying – or, just to get
it in the air is incredibly expensive.

And then to process all of those
data are really expensive as well.

So the bottom line there is
that it is cost-prohibitive.

If you’re sampling a large area,
it’s well worth it, but the frequency

that we want to sample, it’s just not
economically or fiscally responsible.

- Not until the instruments are,
like, pint-sized, right?

- Exactly. Yeah, so and I –
you know, we’ve looked into,

are there small Lidar units
available to put on these drones?

And I think folks are in the process
of trying to develop them.

And actually, there are small Lidar units,
but they’re just – they’re just not that

great right now in terms of
the accuracy that we want.

So right now we’re – yeah, we’re just
plunking down a camera and using that.

And it works really well. So far.

- Thank you.
- Yeah, thanks.

One more?

- Here’s a fun question.
- Uh-oh.

- Have you ever [static sounds]
been in an avalanche or [inaudible]

actually surf out of
an avalanche? [laughs]

- Yeah, so the first part of that
question is, unfortunately, I have been.

And it was, thankfully – well, maybe
I shouldn’t say – it has been twice.

And they were small avalanches.
It was early in my career. [laughter]

So they were relatively small,
and I wasn’t buried, which is good.

And they were – you know, I like to
think that I chose these sort of small,

low-consequence slopes, and
that’s where I was testing things.

And it was actually during –
when I was conducting a stability test,

and it was just right after that.
But, again, they were really small slopes.

So can you surf out of an avalanche?
You know, it really depends.

Sometimes – a lot of it
depends on the debris flow.

And within, again, the past

development in technology called
a balloon pack, or an airbag pack.

And so you basically have a
backpack on, and it has a little ripcord.

And it’s powered by
either a fan or a gas cylinder.

And you can pull this ripcord,
and you basically have a big balloon

that – a big airbag, basically,
that blows up right around your head.

So you have this big
balloon around your head.

And the theory there being is that
these – sort of like the Brazil nut effect,

that you have these – you know,
the Brazil nuts always end up on the

top in your mixed nuts canister.

And so the – you basically get to –
you get basically pushed to the surface

as you get mangled through
the avalanche debris.

So, you know, if you’re not –
if you’re not injured, or if you don’t die

by trauma, then, you know, they do
help with, you know, survival rates.

And they’re not the
silver bullet, by any means.

You know, people have been
fully buried while still wearing

an avalanche balloon pack,
or an airbag pack.

And so they’re not a silver bullet,
but they do help.

If you don’t have one, you know,
the advice from sort of age-old

avalanche hunters is that, you know,
you just got to – if you’re caught in one,

you fight, and you just try
and fight to stay near the surface.

And again, that’s much
easier said than done.

And, yeah, they’re really powerful –
really powerful things.

- Any other questions tonight for
Erich about snow avalanches?

Well, I want to thank you
all for coming and joining us.


And especially I want to thank Erich
Peitzsch for such an enlightening talk.


Thank you.
- Thanks, everyone. I appreciate it.