Water, sediment supply reqs—post-wildfire debris flows in western U.S.

Video Transcript
Download Video
Right-click and save to download

Detailed Description

We sometimes fall back on an implicit model that post-wildfire debris flows are only triggered once a certain level of saturation or rainfall is reached, and that the sediment supply is soon exhausted and future flows are unlikely until it recharges. In this research, we explore a revised model, that there is usually ample water in the system both before and after the flow, that antecedent moisture only plays a small role, and that sediment supply is not a limiting factor (this last one seems to be the case right now in Glenwood Canyon). We will also look at quantitative changes in debris flow abundance and impacts as a result of changing climate.

Santi P (2021). Water and sediment supply requirements for post-wildfire debris flows in the western United States. USGS Landslide Hazards Program Seminar Series, 15 September 2021.


Date Taken:

Length: 00:46:39

Location Taken: US

Video Credits

Video thumbnail: Debris flow path near Glenwood Springs, CO showing naturally formed levees. (Paul Santi, Colorado School of Mines)

Lisa Wald, USGS



So, hi, folks. My name is Matt Thomas,
and I just want to thank you for

tuning in to the USGS Landslide
Hazards Program seminar series.

For those of you that are new to this
meeting, you have the ability to submit

questions via the chat window or to
use the raise-your-hand feature

in combination with your
microphone and video camera.

We typically wait until the end of
the presentation to take questions,

but you can enter them into the
chat window ahead of time so that

your questions are waiting to be
answered at the conclusion of the talk.

In the meantime, please just do your
best to make sure your microphone

is muted and your video is off
when you aren’t intending to speak.

Thanks. I’ll let you
take it away, Francis.

- Well, hi, everyone.

I’m really excited to be able to
introduce Dr. Paul Santi today.

You know, Paul got his start at
Duke University, where he

majored in geology and physics.
And then he went on to Texas A&M

and got into the practical side of things,
so with geologic engineering,

and went to the Bay Area, and –
which is sort of the mecca

for landsliding and, you know,
kind of geotechnical problems.

And really spent a good amount of
time getting a sense of, you know,

the wide variety of geological problems
that people face on a practical level.

He came back to Colorado
School of Mines to do a Ph.D.

And then, from there,
went on to Missouri S&T,

or, at the time,
it was Missouri-Rolla.

Taught there for a few years and
then Colorado drew him back to here.

And so he’s – came back
to Mines to be a professor.

He has been the president of both
AEG and GSA – the GSA

Environmental & Engineering Geology
Division. He’s also a GSA fellow.

Recently, Paul has started the
Center for Mining Sustainability.

This has taken him to Peru
pretty frequently.

And he’s gotten good enough that
he’s actually even giving talks

in Spanish now, which is really –
but the really important thing

to know about Paul is that
he’s an Ironman triathlete.

- [laughs]
- And he has a really, really

high endurance level.
So, not only is he a good scientist,

but he can outlast just about any of us.
So, anyway, with that introduction,

I’ll turn it over to Paul.
- Thanks, Francis.

I’m a lot older than that now.
I have zero endurance,

so I’m not going to challenge anybody.
But, you know, thanks for the

opportunity to come and talk to
this group. This is a great group.

I’ve had to miss a lot of the talks because
I often teach on Wednesday afternoons.

But I love the lineup, and
I’m proud to be part of that.

And so what I want to talk about
today – I’m going to focus mostly

on really a couple hypotheses that
we explored and wanted to support

regarding water and sediment supply
for post-wildfire debris flows.

And there’s a couple questions
that we’re trying to answer.

I’ll also talk a little bit about kind of
what things look like for the future

and some modeling we did
looking at climate changes.

And, by “we,” I mean other people –
[laughs] these two students that are

listed here – Blaire MacAulay,
who got her master’s a number of years

back and works for Baseline Water
in Calgary, and then Zane White,

who just finished in
August with his degree.

And so I’m presenting their work.
I was their adviser, which means

I had some opinions, but I’ll say that
they did the lion’s share of the work.

And, if you have any really
difficult questions, you’re going to

have to ask them.
So I can ditch things off to them.

And so, really what we’re after here is,
I want to – I want to sort of adjust

at least how I think about debris
flows, and maybe how all of us.

And I think there’s an implicit model
that we don’t often state, but it’s this

idea that a debris flow is triggered,
that we start out with a dry situation,

dry soil. It starts raining. Once we have
enough rain, a debris flow lets go.

And I think that’s true for debris flow
intensity, but I want to kind of

demonstrate that there’s more
than enough water in the system.

And then the second part of the
implicit model is that there’s a –

there’s a finite supply of debris and
this idea that a debris flow kind of

cleans up a channel, and you have to
have some time for the gun to reload.

And this is – actually, I didn’t use those
exact words, but it’s an implication that

I made that I did my master’s thesis
in Utah, like, over 30 years ago.

And it was not a
fire-related debris flow.

That wasn’t quite
as much of an issue then.

But it ended up being an important
point, and I would get a call every few

years from somebody in Davis County,
Utah, saying, well, you know,

there was a debris flow in this channel.
Do we need to worry about it,

or is – or is there no supply left?
And we’ve seen a challenge to this

just in Glenwood Canyon
the last couple months

when we get repeat
flows in the same area.

And, you know, there’s this sense,
like, when will it end.

Every time it rains, there’s
enough debris for this.

So that’s sort of our implicit model.
And I think we can tweak it a little bit

and I think it will help us when we
consider the hazards for post-wildfire

debris flows to say, first of all,
there’s ample water before and after

the debris flow.
And there’s – it starts raining.

There’s water running through the
system. Then there’s a debris flow.

Then there’s more water
running through the system.

And we’re not waiting to have
enough overall to cause an event.

But we are waiting to have
this sort of intensity burst.

And then – and then the second point of
that is that the sediment supply is ample,

and it can produce multiple events.
And I would guess that most people

that deal with post-wildfire debris flows
probably agree with this hypothesized

model and have kind of said a lot of
these things over the years themselves.

And we’re just trying to provide some
backup for that and some data to say,

yes, here is what you
can rely on to show these.

So where does our data come from?
The rainfall data comes from a series

of debris flows. There are 44 of them
that occurred in nine different

burned areas, mostly about
15 or more years ago, in Colorado,

California, and Utah.
A range of drainage basin size.

The average is about a square kilometer.
So these are not huge basins.

But there’s a few that are very small
and a few that are very large.

And then, for each of these 44 events,
we know the debris flow volume, and

we had some detailed rain gauge data.
And the rain gauges – some were

right there in the same watershed.
Some were nearby.

Maybe we had to do an inverse distance
averaging of a few different gauges.

There are obviously changes in
rainfall with elevation and all that.

So, you know, you have to kind of
squint to use the rainfall data, as always,

if you’re going to use it to represent
what’s happening over an entire basin.

And then, for sediment supply,
the data that we used is distributed

over the Pacific Northwest
and the western U.S.

Had a pretty large database of events
broken into three categories –

recently burned areas; areas that are
sort of actively recovering from

a burn – so that means within, let’s say,
1 to 10 years after a wildfire; and then

some unburned areas, or areas that have
had more than 10 years since a wildfire.

So then the model that we looked at
was really two systems.

And we said, okay, first of all,
in some of these drainage basins,

where there were post-wildfire debris
flows, there is unburned ground.

And so maybe the fire only covers 60%
of the basin or something like that.

So we’re calling that unburned area.
We’re only worried about the soil

and saying there’s some
absorptive capacity for the soil.

We’re not really worrying about
water percolating through the soil

and into bedrock and considering
bedrock as part of the model.

And, on short timeframe, like one
rainstorm, it probably makes sense

just to consider the soil portion of it.
and then, in the burned area, we’re

saying, yes, there’s ash on top of it.
And there’s – and that’s sort of a –

there’s a permeability contrast
between the ash and the soil,

and so we’re worried about how
much water that ash can absorb.

We also made assumptions
of horizontal topography.

Even though these are steep places that
might have less soil where it’s steep,

but that the conservative assumption is
to say the water velocity and all of that,

we’re going to ignore.
We’re just going to say, water hits this,

and it can either absorb or not on the
basis as if it was flat topography.

The ash thickness –
we assumed about 10 millimeters.

And this is – there are some
other people that have modeled

this same thing. There are some
people that have done some field

measurements. We felt that was a
conservation, but realistic, estimate.

And then, for the porosity,
as the absorptive capacity of the ash,

we assumed around 20 to 30%.
And people have measured much

higher, and that’s usually almost
immediately after the wildfire.

But it compresses pretty quickly
on its own, or, if you get a little bit

of water in it, it compresses.
And so, again, we wanted to have that

sort of realistic sort of average value.
So we used 20 to 30%.

Then we assumed, when water hits
the ash, that there’s 100% saturation.

That there’s – that there’s no transfer
of water from the ash to the soil

because of this permeability contrast.
And then, once the ash is saturated,

there’s runoff. And so this is the
fill-and-spill model that several people

here have published papers on.
And, again, that’s going to be

our assumption for
this water balance.

And then there’s this idea
of antecedent moisture.

How important is it, whether the
soil has some moisture content

or preceding rainfall.
And we didn’t consider this,

and I’ll address it later in the paper –
or, in the presentation.

I think we’ve got some evidence that
it’s not a real strong factor here.

And then finally, for unburned soil –
and we’ll get into this –

but we’re assuming that the
infiltration can be modeled

using some soil conservation
service – NRCS runoff curves.

So what that leads to, then,
is water balance equation.

And this is really what we’re after.
And so, on one side – on the left-hand

side is the input to the system.
And that’s the rainfall.

And really, we’re looking
at a volume of water.

How much storm rainfall was there
times the drainage basin area.

And so the total rainfall over this
period of time that caused the debris

flow times the drainage basin area.
So that’s the input.

And then, on the right-hand side is
all the places that water can go.

And we said, really,
there’s three groups.

The first is, it could be incorporated into
a debris flow. And we can measure that.

It can infiltrate into the soil or the ash.
And then the remainder – the leftover –

is the overland flow. And we wanted
to look at the balance of this.

And, for the hypothesis that I’m
proposing here, I’m saying that,

once you account for the infiltration,
once you account for the debris flow

water, that there is a lot of water
left over for overland flow.

And that fits the model that says,
it’s raining, there’s water running

through the system,
then you get a debris flow,

and there’s still water
running through the system.

So this is how we would estimate
rainfall based on nearby rain gauges

and so on – size of the basin.
Debris flow water, we said, okay,

we measured the volume of debris
in a number of different events –

44 different events.
And we have to make

an assumption of how much water was
incorporated into that volume of debris.

And we’ve used some rules of thumb
that people have used over the years

and said, it ranges between 20 and 40%.
And the 20% says, we’re not using

very much water, so the leftover
that’s the overland flow would be

sort of the high end of the bracket.
And 40% water – that lower row –

would say, that’s the maximum
amount of water in the debris flow,

so that’s the low, or minimum,
overland flow.

And so we’re using those
to bracket the ranges.

Ash infiltration – again, we’re saying
there’s a porosity range in the ash.

There’s a high and low porosity range.
The thickness of 10 centimeters

times the burned area
within that drainage basin.

And so that assumes that any place
that burned and was marked on the

maps as burned has ash present.
And any place that was not burned,

the ash is missing. And obviously,
there’s going to be some different

situations there. The ash thickness
might not be consistent, but we said,

you know, just for practicality, we’re
going to make these two assumptions.

Unburned soil infiltration
is a little more complicated.

But, again, we went to the NRCS data.
For those of you that understand it,

or have used it, we assumed an
Antecedent Moisture Condition II,

which is not bone dry,
but it’s not saturated.

And we said it’s fair to say there’s
a few percent moisture content.

Maybe not the day after the fire,
but within a few weeks

or when most of these
debris flows occurred.

We picked a range of hydrologic
soil groups to cover this area.

And these are – they’re a little different,
depending on whether we were

dealing with slides in southern –
or, debris flows in southern California

or Utah or Colorado.
So we sort of matched it to the climate.

But it shows a pretty wide range
of textures and infiltration rates.

And, again, we want to say,
is this a problem if you

make an assumption that there’s
a broad range of properties.

And then finally, for hydrologic
conditions, we took a range from –

again, for unburned, a pretty
low percent ground cover

to sort of a medium percent
ground cover, which makes sense

for the drier part
of the western U.S.

And then, basically, what this allows
you to do then is select curve numbers.

From curve numbers, there’s
a basic equation that you can use

to calculate what the potential
retention in the soil is.

And then, having that number, you can
calculate what the potential runoff is.

And so that P-sub-e
is the potential runoff.

That’s the number that
we’re including in our equation.

And then, leftover from
this is overland flow.

So that’s the unknown
that we want to look at.

So the results for this, then.
Here’s a set of diagrams.

And this is for the low flow.
So this says we’re putting as much

water as possible in the debris flow,
and the remainder is the overland flow.

The three colors in the graph, then –
the bottom bar – the blue one is

how much water from a rainstorm –
the percentage of water at that basin

that would incorporated
into the debris flow.

The red bar in the middle is the
percentage of water that would infiltrate

under sort of a maximum
infiltration condition.

And then the
reminder is the green.

And so what you notice here is there’s
only about – there’s only four of these

that really wouldn’t have any
runoff under these conditions.

There’s a few more that would
have very little runoff.

But the vast majority of these,
there’s a lot of green on that graph.

And so that says you can – you can
look at how much water the debris flow

would incorporate compared to
how much water came out of

this rainstorm over, you know,
typically a couple-hour period,

but sometimes
it’s multiple hours.

So that’s the low flow.
Here’s the high flow.

And so this says we’re not putting
as much water into the debris flow.

We’re not infiltrating as much.

And you can see there’s
even more green.

And so I’ll toggle back
and forth between these,

and you can kind of get
a sense of what the range is.

But, to me, this – every now and then,
data works the way that you want it to.

[laughs] I looked at this, and we plot
graph, and it’s, like, that’s great.

To me, that makes a really strong point
that there’s excess water in the system.

And so another way to look at that,
we said, instead of 20 to 40%

of the debris flow being water,
let’s pick the mid-range. It’s 30%.

And I want to know how
much of the rainfall is incorporated

into the debris flow.
And that’s what this bar chart shows.

And the median is 5.7%,
which is 1/17th of all the water,

is incorporated into the debris flow.
And so, again, I think that supports

a model that says, it’s raining,
there’s water running through

the system, then there’s a debris flow,
and then there’s more water.

So we’re not limited
by the amount of water.

We’re not waiting for enough water
in the system before this happens.

And so this idea of rainfall intensity –
this is – this is not really a new concept.

But I wanted to show some data
we have from the Great Sand Dunes

National Park.
Basically to show how important

intensity of rainfall is as opposed to
the total amount of rainfall.

And so the upper two graphs are
from two different drainage basins.

We put pressure
transducers in the channels.

And so we could record, then,
there’s a little bit of pressure

as water is passing through it
during a storm, and then it jumps up

pretty high when the
debris flow comes over.

So we really have, within about –
I think we were on a 15-second interval,

so we know, you know,
within less than a minute

of when the debris flow occurs.
And that’s the red arrows in the timing.

The bottom two graphs
are rainfall intensity.

And so the rain gauge for Basin 24,
I think, was 10 meters away

from the transducer.
And so this is almost an exact,

you know, measurement of
rainfall versus debris flow.

Basin 16 was about –
it was less than a kilometer

from the transducer in Basin 32 –
about the same elevation.

So there weren’t really
orographic issues.

But what you can see is that
Basin 24, there’s, you know,

a few little peaks in rainfall.
And then there’s – in the middle of that

graph, there’s sort of a larger peak.
And that thing ends.

And within maybe two to three minutes,
you get a debris flow at 5:10.

And, on the right-hand side, those
two graphs, there’s two debris flows,

and you can really see the correlation
between the rainfall intensity peak and,

again, two or three minutes later,
there’s the debris flow response.

Like I said, we’re not
the only person to do this.

Here’s some data that
Sue Cannon had from years ago.

These are cumulative rainfall graphs
instead of just incremental intensity.

So, wherever the graph is steep means
you have a increase in rainfall intensity.

And then the blue triangles are
the timing of the debris flow

that came through. And so, again,
you can see a really close link.

I picked three. She had –
in this study, I think there were

about 10 different ones.
I know other people at the USGS

have measured the same sort of thing,
and I think outside the GS as well.

And then one other interesting
thing with rainfall intensity.

And, if you’ve worked with debris
flows, you’ve seen these graphs –

intensity duration threshold graphs.
And so each line here represents

a series of measurements at a different
drainage basin or different area.

And so what someone does in that
area is you’re measuring rainfall,

and sometimes there are rainfall
events that don’t cause debris flows.

And other times, there are rainfall
events that do cause debris flows.

And so, for those two different groups,
you can separate out some different

duration intensity pairs
and plot those on the graph.

And so, on the upper part of your graph,
you have a whole bunch of X’s,

where you say, these combinations
caused a debris flow.

And, on the lower part of your graph,
you have a bunch of circles,

and you say these combination
did not cause a debris flow.

And you draw a line
in between them,

and that’s the duration
intensity threshold.

And so there was a – these have been
accumulated by people over the years.

And then Sue Cannon and Jerry
DeGraff published this version where

the gray lines are the thresholds
for non-fire-related debris flows.

And the colored lines are thresholds
for fire-related debris flows.

And what you can see is that
it requires a lot less rain to cause

a debris flow in a burned area.
And those lines are a lot lower.

So the intensity-duration
pairs are much lower.

Again, it’s nothing new.
But, to me, it shows that there’s a

rainfall intensity link that’s easier to
achieve the threshold after a wildfire.

And then I’ve added
a couple lines on here.

The black line is what we measured
at the Great Sand Dunes.

And there were – I can’t remember –
I think maybe four different basins

that had debris flows.
And so we had a number of storms.

And, again, we put in a pretty solid line.
It’s right there within the

group of burned basins.
Great Sand Dunes was a little unique.

Even though we were up in the
mountains, there was a high component

of windblown fine sand that had come
from the dunes up onto the residual soil.

And so that changes the character
and the – and the kind of erodibility

and the – and the hydrologic
response of the – of the soil up there.

And so that’s probably why
that line is a little lower than

kind of the average
threshold for burned areas.

And then the black dashed line
was our – basically the threshold

that we established over the
next two years of measuring.

And I make it a dash because
no debris flows occurred

during the next two years.
So the first years after the wildfire,

there were at least four – I can’t
remember the exact number.

And we drew
the solid black line.

And then, for the next two years,
we monitored this.

And the watershed – or, the various
watersheds had really recovered.

And there were no debris flows,
even though we were getting

stronger rainstorm events.
And so that dashed line kind of

represents the maximum.
And so we know the true threshold

is at least that high,
and it might be higher.

But it was really cool to see this
much recovery and change

in sensitivity to rainfall intensity,
just within a year or two.

So, as I promised [chuckles],
I’ll talk a little about antecedent

moisture and how important
is rainfall ahead of time in causing –

in allowing a debris flow to occur.
And so what the data shows here –

the left-hand column,
I’ve basically listed the names

of several different fires that
we were working with and

their number of drainage
basins for each fire.

And then I have a number 1,
number 2, number 3.

Those are different weather stations.
And so we tried to sort of bracket this

fire by picking weather stations at –
that were obviously nearby or

within some of the drainage basins,
but at different elevations so that

we could account for any orographic
changes in the rainstorms.

So, for instance, the first few rows – the
Old and Grand Prix Fire in California.

It’s a pretty big area. We’re looking
at all the drainage basins,

and we used three different
weather stations. And so on.

And so then the second column –
the middle column just says,

you know, which drainage basins
that had debris flows are we –

are we talking about –
are we comparing this to.

Then finally, the right-hand
column is really the meat

of the information,
and there’s two parts of that.

There’s how much rainfall
occurred in the entire previous week

at that weather station.
How much occurred in

the entire previous month.
And these are millimeters.

And so what you can see, in the
previous week, the most we had was

5.6 millimeters, except for the Mollie
Fire in Utah that had a little bit higher.

So, you know, again, millimeters –
it’s not a huge amount of rainfall.

But, for the vast majority of these,
this is – you know, half a centimeter

within a week, that’s not saturating
all the ash and all the soil.

And, if you look at the previous month,
I highlighted in red anything that

was greater than 50. And, again,
for most of these, it’s less than 50.

That’s – you know, that’s 5 centimeters
in a month spread out.

There have been some studies that
show that it takes a month or two

for the system to kind of
re-equilibrate after rainfall.

And so you could have
some residual effects here.

But I think – to me, this shows pretty
strongly that you don’t need

a wet week beforehand and then
a storm that causes a debris flow.

You can – you can have pretty much
bone-dry soil that has had very little

rainfall – almost none in a week, and
very little within the previous month,

and you can still get a debris flow
that occurs during the rainstorm.

Next I want to move on and talk a little
about sediment supply abundance.

And I’ve got a few different – I’d say
nothing is 100% diagnostic, but I think

there are a few different things that –
clues that sort of point to the idea that

there’s plenty of sediment in the system
for these – for these debris flows.

And we’re not running out of sediment.
We’re not limited by the amount

of sediment. And so this first group,
what I want to show you –

there are two graphs here that
I’ve been carting around for years

because I love them.

And the upper one –
western U.S. – WUS less than 1.

So that’s recently burned areas in the
western U.S. – less than one year.

And the X axis shows how many
different data points we have –

or, the Y axis. And the X axis,
we’re measuring debris flow volume,

but we’re normalizing
by basin area.

So cubic meters of debris
per square kilometer of basin.

Because we know that bigger basins can
probably produce bigger debris flows,

and we wanted to just –
to normalize for that.

And then the solid black line –
vertical line shows the median value.

So western U.S., less than 1,
median value is 7,188 cubic meters

per square kilometer. Then the lower
graph is western U.S. greater than 10.

So that’s areas that have –
either have not burned or

they’ve had at least
10 years of recovery.

And the median value is much less.
And so the – over on the left there,

I’m saying sort of how
many events are in each group.

And then really what we found –
that second set of bullets is that,

when you normalize it for basin size,
the data I’m showing here, debris flows

are more than twice as big after
a wildfire within that first year

as compared to unburned areas.
And we’re ignoring differences in

rainstorm and all these other things.
We’re just saying we’ll look at

two big data sets.
Is it worse in burned areas?

And assume the data sets are
big enough that they’re covering

sort of the same range of conditions.
And so the answer – it is worse.

And so kind of comparing apples
to apples and using this normalized

for basin area. And then we used
a cluster analysis to try to do

even more normalization.
And so that says we normalize

for basin size, and then we took
smaller groups that had similar

channel length and basin gradient.
So now we’re comparing, like,

Granny Smith apples to
Granny Smith apples.

And, in that case,
there’s even a bigger difference.

And I’m not showing the graphs here,
but, instead of roughly twice the size,

burned debris flows are
2.7 to 5.4 times bigger.

And so, to me, what this says is that,
in a burned area at least,

there’s, first of all,
a lot of sediment in the system.

Second sort of line of evidence is
looking where the material comes from.

And so we had a series of graphs –
this is – this is one of them.

We had, I think, 54 of these, where
I had grad students over a couple

summers basically walk up the burn
channel and measure cross-sections.

And so you can see, on this –
on the right-hand photograph,

that – you can see how much the
debris flow has scoured out material.

And basically, by measuring a cross-
section across that, and then you can

make some assumptions as to what the
pre-debris flow topography looked like.

And you can see where it’s cut off
there and where the grass ends.

And so we could compare that to
adjacent channels that had not failed.

So we knew what
the shape should be.

And so we could estimate, then, what’s
the area that has been scoured out,

and multiply that by the distance sort of
average to the next cross-section.

And so we can have an incremental
volume between these cross-sections.

And so what this graph shows – every
square there represents a cross-section.

And you can see from the top of the
channel, where it says distance down

channel is zero, and then
to the mouth of the channel.

You can see the growth
of the debris flow and

how much material
is coming from the channel.

And we also looked at evidence of rills,
of smaller gullies, of sheet wash.

And, based on some measurements
and estimations, on average,

we concluded that about 3% of the
total volume was coming from

those sources as opposed to the channel.
And there’s some that – some basins

that were larger than that
and some that were smaller.

But the median value was 3%.
And so then, to me, this says there’s

a lot of material in the
channels being eroded.

Sometimes it’s eroded to bedrock,
but, like in this case here, it removed

a lot of material, but there’s still
a lot of sediment in the channel.

And then, finally,
you get bank failures.

Like – this might be
Rich Giraud from the

Utah Geological Survey
that spoke last week.

But you can – you can see that it
over-steepens the banks, and then,

for a period of time, you’re recharging
this thing by bank failure.

And then the third line of evidence
I would say for sediment supply

abundance is this idea that
you can measure multiple

debris flows in
the same canyon.

And particularly this figure on the left –
you can see how it’s eroded material

and removed it. But there’s
still a lot left in there.

Same thing on the right.
There’s some bank failures.

Again, this is – this is what we’re
seeing in Glenwood Springs right now,

in Glenwood Canyon, is multiple
events in the same canyon.

And so I took two
different data groups.

The top table is from the
Coal Seam Fire, 2002,

and the Missionary Ridge Fire is
the bottom table. That’s also 2002.

And basically said, for these different
watersheds – so each row is a different

watershed, when were
there debris flows occurring.

And I just put an X when there
was a measured debris flow.

And what you can see, then,
for the Coal Seam Fire,

six of these nine basins
had multiple flows.

In fact, as many as four or five flows,
just in that first season.

Missionary Ridge Fire –
15 of the 16 basins had multiple.

And, again,
as many as four flows.

And so it just – you have a debris flow.
A few weeks later, you have

another one. A few weeks later,
you have another one.

And, at least in this first
period of time, it appears like

there’s an infinite
source of sediment.

Now, obviously, we’ll clean things out
of the system over a long time period.

But, for a short-term hazard, for
the first couple years while you’re

still really sensitive to, you know,
the rainfall and all of that, you just have

to assume that there’s enough debris
available to cause a debris flow.

So that kind of concludes
those two hypotheses.

I wanted to move on to the work that
Zane did recently and says, okay,

we recognize this problem is bigger.
It’s really sensitive to rainfall intensity.

There’s more than enough water.
There’s more than enough sediment.

What does this mean?
What does this bode for the future?

And so we looked at the Thomas Fire,
which occurred in 2017.

You may recall, in January of 2018,
there were this horrendous

number of debris flows.
They flooded Oprah’s backyard.

This was the Montecito,
California, events.

And I say the size of the fire there –
it caused – the debris flows caused

23 deaths, 200 million in damage.
This grabbed the public’s attention.

And so we said, well, what would
happen if this flow was – if this event

was to occur in 2050 or 2075 – in the
future, based on climate change.

And Jason Kean and Dennis Staley had
a publication last year that was really

nice talking about some climatic effects.
And so we wanted to do something

a little more – sort of parallel,
but more on a very narrow

geographic scale and look at this.
And so we’re going to say,

okay, what would happen if this
burned in 2050 or 2075? What’s the

difference in rainfall? And what’s
the difference in the debris impact?

And so, for – you know, as far as sort of
looking at the hazard and measuring the

impact, we used the things that are in
the emergency assessment page that

the GS publishes and keeps updated.
And there’s two sets of equations.

The first is
likelihood modeling.

And it says, how likely is it to have a
debris flow. That’s part of the hazard.

And basically, there’s a bunch
of variables in there.

This is all empirical.

And I colored in red the things that
would change with a changing climate.

And so the proportion of the –
of the size of the burned area.

The peak 15-minute rainfall
is going to change somewhat.

If you look at the volume modeling –
that second group of equations, there’s –

the area burned can change somewhat.
The rainfall intensity.

And so these are –
these are factors that

we’re going to plug
into this same model.

And then we wanted to say, well,
you know, what’s the climate model

for the future? And there are
a lot of different approaches.

Basically, what we wanted
to use is these representative

pathways – RCPs.

And the two sort of end members that
we looked at is the RCP 8.5 that says,

we’re not going to do anything.
We’re going to let carbon accumulate.

We’re going to let the emissions
continue, not change the climate –

this is the worst-case scenario,
which kind of seems where

we’re headed [chuckles]
these days.

And then we really thought
the best case is the RCP 4.5

that has some moderate
emissions mitigation.

And so we’ll follow these models,
and as you’ll see, there’s not

a huge amount of difference in
the response. It’s definitely visible.

But we’re going to have to do better
than a 4.5, I think, for a lot of reasons,

but also because of post-wildfire
debris flow problems.

And so a couple
things are changing.

I mean, obviously, as the climate is
changing, the temperature is going up.

And that’s going to cause a change
in the hydrologic setting.

And really, the sort of general
conclusion is that the climate’s drier,

but you will have – you will
have more intense events.

And so what we’re comparing this to is,
at least for this location, the USGS used

a designed storm of a 15-minute,
40-millimeter-per-hour storm.

And then we used some equations that
basically said, okay, for every degree

increase in temperature – in average
temperature, you would see intense

rainstorms increasing by 6 to 7%.
And this is a really tricky thing

to model, but it just says,
40% – or, this 40-millimeter

designed storm,
how will that change over time?

And so we used, then, the – I’m not going
to try to – no, I will try to say it –

the Clausius-Clapeyron relation.
And then we used a Cal-Adapt

website that does some temperature
projections for Ventura County.

And so, then, in the tables on the right –
the upper table, you can see the

RCP 4.5 temperature increase
over time, you know, 2 degrees,

and by 2050, 2.4 in 2075. Little bit
more than that with the RCP 8.5.

These are – these are
big temperature changes.

And then, again, 6 to 7% change
in rainfall for each of those –

that’s the bottom table.
And so our 40-millimeter

designed storm now goes
up to 46 to 50 millimeters.

So that’s one thing we expect to change.
And then the fire is going to get bigger.

And people have suggested this in other
publications that I referenced there.

Some of their changes we’ve plotted
here in the different-colored lines.

And we picked a median value,
which, if you can see, it’s the –

it’s the gray line that’s
right below the orange one.

And so it says, by the year 2050, fire
size would be expected to increase 54%.

And, by the year 2075, the fire size
would be expected to increase 81%.

So a typical fire is going to be bigger.
We’ve been seeing this, right?

Three of the largest –
I don’t remember –

three of the largest 10 fires
in Colorado occurred last year.

And so we’re seeing
these bigger events.

So, because we have a bigger fire
boundary, we want to know, what are

the drainage basins. And so we used
this LANDFIRE system to do that.

We only dealt with the larger
basins bigger than 2 hectares.

Interesting to note, then,
that the current fire has,

the way we calculate it,
1,736 watersheds.

If you increase it by
year 2050 to 154%, look at this.

We have over 3,000 basins –
3,115 basins.

And then, increasing it even more
by the year 2075, 3,500 basins.

So there’s – just from that standpoint,
there’s a lot more potential for

debris flows because there’s,
like, almost – there are more than

double the number of
basins that have burned.

So, using sort of our methodology
and our drainage basins,

we reproduced some of the USGS
models. You can pull up their website

right now and compare them.
They’re pretty close to similar.

But there’s the likelihood model.
In this case, brown is the worst.

There’s a volume model.

And you basically group those by
intensity and produce a hazard model.

And so this is what we’re going to
be working with is the hazards.

It says this basin has the highest hazard.
The red ones have moderate, and the

yellow ones have lower hazard.
So that’s our comparison piece.

So, if we just look at change in rainfall,
just say, okay, we’re going from

40 millimeters to 46 to 50 millimeters,
the left-hand graph shows the –

basically what happened in 2017.
The middle figure there shows

what we would expect by 2050.
And you can see there’s

more brown
and little bit less red.

And then the lower right-hand
graph says this is what we would

expect by the year 2075. And you
can see there’s even more brown.

These aren’t really
dramatic changes.

But you can see there’s a shift towards
higher hazard just within that.

And then we want to add on top
of that the change in fire size.

And so this shows, in blue,
the 2017 fire boundary.

And then, if we expand it
by 54%, that’s the yellow.

And if we expand it even more –
so this is kind of rubber sheet

expansion – if we expand it 181%,
that’s the year 2075,

if the fire were to occur then,
that’s the red boundary.

Couple things to note.
On that sort of southwest side

is the ocean, and so it doesn’t
expand in that direction.

There’s an unburned island in the
middle that’s based on the development

and everything. We more or less
retain that the same way it was.

But everything else
stretches very evenly.

And we also did the same thing with
fire intensity, is taking the current fire

intensity map and stretching it.
And so it monkeys with the basins

a little bit, but it’s easier than
doing this individually

for 1,700 basins
or 3,400 basins.

So the results,
then, from this.

So this is – this is rainfall
changes plus fire size increase.

And you can see, again, the left-hand –
lower left is basically 2017.

The upper-middle graph
is the year 2050.

And the right-hand
graph is the year 2075.

They look similar because,
when we stretch this thing out,

it has, like – the boundaries
are sort of mirror image.

But, if you look really closely at some
of the features outside, you can see

how this thing is getting bigger and
it’s incorporating more basins.

You can summarize some of that.
I don’t want to spend a huge amount

of time here, but basically what
this shows – the rows – the top row

is the likelihood. How likely
is it to have a debris flow.

And these are divided by,
like, 20% ranges.

So 5 represents the 80 to 100% chance
of getting a debris flow, and so on.

So the top row is likelihood.
The second row down is the

volume classification.
Third row down is adding those two.

So that’s sort of the
hazard classification.

And then the bottom row says,
we’re going to take the hazard

classification and sort of
renumber it into three groups.

And then each column
shows a different rainstorm.

So blue – the left-hand column
shows what happened in 2017.

The middle columns are the year 2050.
And then the right-hand column

in red is the year 2075.
And the point I want to make

with this is you can see a shift
towards the more hazardous zones.

So there’s more basins,
but there’s –

it’s more likely to have
higher-hazard basins.

And then finally, I want to show
this comparison of the models.

This is a little bit busy table,
but I’ve outlined in a beautiful

olive green what I think
are the important rows.

And so the top row –
2017, 40 millimeters, 100% –

that’s normal fire size.
1,736 basins – we’ve seen that.

And here’s the total volume of debris
that would be calculated using this

model. 11 million cubic meters.
And 54% of the basins are high hazard.

So the next outlined
basin I have is the 2050.

You can see higher rain intensity,
bigger fire size, 3,100 basins now,

twice the debris volume, 60%
high-hazard basins instead of 54%.

And then, if we continue on these
climate models and look at the

year 2050, the fire’s a lot bigger.
There’s more than twice as many

basins – 3,559 basins.
There’s 28 million cubic meters

of debris in this system.
And you could scale that up

from a damage and
dollar volume as well.

And more of the basins
are high-hazard basins.

So, to summarize those results then,
by the year 2050, there’s 75%

more burned basins. By the year 2075,
it’s more than double. It’s 105% more.

The volume is a lot larger.

And the percentage of high-hazard
basins is a lot higher.

So then, to conclude, putting all
this together, even with pretty

conservative assumptions,
there’s excess water in the system.

Debris flow doesn’t so much
depend on reaching, you know,

this magical amount of water,
but it depends on this intensity.

And so the limiting factor for forming
these is the dynamics of the pulse of

water and not
the sediment supply.

And then climate changes are going to
multiply this hazard significantly.