PubTalk 5/2018 — Yes Humans really are causing induced earthquakes

Video Transcript
Download Video
Right-click and save to download

Detailed Description

Title: Yes, Humans Really Are Causing Earthquakes! How Energy Industry Practices are Causing Earthquakes in America's Heartland

  • In every year since 2014, Oklahoma has had more earthquakes than California.
  • Oil and gas operations are "inducing" these earthquakes.
  • The earthquake rate has dropped by more than 50 percent due to changes in industry practices.


Date Taken:

Length: 01:21:32

Location Taken: Menlo Park, CA, US



Good evening.

Oh, good. I get response.
[laughs] That’s great.

Welcome to the USGS
evening public lecture for May.

And we’re just sneaking this in
the last day of the month.

Mitch? Is it on? Okay. They told me
to hold it closer. Thank you.

Due to some unforeseen circumstances,
we will not be having a June lecture,

but the good news is, we will resume
and have a July lecture on July 26th.

And it’s going to be
on acid mine drainage.

So I’m really encouraging you all to
pick up the flier in the back – trying to

fumble with this – so that you can
remember July 26th for that lecture.

But what you’re really here for tonight
is our lecture, Yes, Humans Really Are

Causing Earthquakes – How Energy
Industry Practices are Causing

Earthquakes in America’s Heartland.
And it’s presented by Justin Rubinstein.

Dr. Justin Rubinstein is a
seismologist and deputy chief

of the Induced Seismicity Project
here at USGS in Menlo Park.

His research focuses on the
ongoing surge of seismicity in the

central United States and its
relationship to oil and gas operations.

This work includes developing
methods to estimate the likelihood

of earthquakes induced by oil
and gas operations and field studies

of seismicity in Colorado,
Kansas, and Oklahoma.

Dr. Rubinstein has worked on
many topics related to earthquakes,

including earthquakes forecasting,
controls on earthquake ground shaking,

and causes of damage in the

near Los Angeles.
Justin received his bachelor’s degree

from UCLA and his master’s and
doctorate from Stanford University.

So without further ado,
let’s give a warm welcome to Justin.


- Thanks, Diane.
So, as you can see from the title

of this presentation,
I’m pretty definitive about

whether these earthquakes are manmade.
We are causing these earthquakes.

And what you can see here is
just a picture of a pumpjack

on the left-hand side and a count of –
annual count of earthquakes

in the central
United States by year.

You can see that, for the –
from 1990 through 2008 or so,

the earthquake rate was
more or less constant –

somewhere around 20
magnitude 3 earthquakes per year.

And something dramatically
started changing in 2009,

such that we had a spike of over

Fortunately, it has begun to decline
in the last couple of years.

And most of these earthquakes
are actually in Oklahoma.

I think most of us think of tornadoes
when we think of Oklahoma.

But I just found this Photoshop of a
Welcome to Oklahoma sign, and it’s

the home of not just tornadoes, but also
earthquakes, so the quakenado is there.

So, yes, humans are causing earthquakes,
but it’s not in the way that you think.

It’s not Max Zorin trying to put nuclear
bombs in the San Andreas Fault.

[laughter] Nor do we have James
Bond to save us in A View to a Kill.

Similarly, it is not Lex Luthor trying
to flood the San Andreas Fault

in Superman I. And, again, we do
not have Superman to save us, either.

Remarkably, though, the way that
Lex Luthor is trying to cause these

earthquakes in this movie in 1979
is actually the way these earthquakes

are being caused, by injecting fluids
deep underground and into faults.

So I apologize we don’t have any
superheroes to protect you from these

earthquakes. You’re just left with
scientists and regulators. [laughter]

So what we’ll do over the course of
this talk is, I first want to give you

a little bit of a history on induced
seismicity – these manmade earthquakes.

And it’s actually much longer
than you probably think.

Then we’ll go into how these
earthquakes are actually physically

caused and the oil and gas
processes that cause them.

And we’ll go through a couple of
case studies and sort of broader studies

of induced seismicity –
some of the more recent work.

And finally, we’ll wrap up
and start thinking about

where do we go from here?
What is the outlook?

Yes, the earthquake rate is going down,
but is this something really permanent?

So, as I mentioned, induced earthquakes
have been occurring for over 100 years.

Now, just a brief stop for terminology.
When I say “induced,”

I mean an earthquake that is
caused by human activity.

It’s an earthquake that probably
would not have occurred if we

weren’t doing something.
I may also use the word “triggered.”

I use these words interchangeably.
I use them to mean the same thing –

earthquakes that were
caused by human activity.

So the first known induced
earthquakes occurred in 1894,

and they were felt in Johannesburg.
Johannesburg, at the time,

was a very small city.
It was a gold-mining city,

and they were actually mining gold
directly underneath the town.

And they were mining in what’s
called room-and-pillar-style mining.

And these are – it’s exactly
what it sounds like.

You carve out a room and
leave just small pillars.

Think of what a parking
structure might look like now.

And so what was happening is, you
would start seeing collapses of these,

and you would start feeling
the shaking at the surface.

And so that’s what these first
human-induced earthquakes were.

And mining-induced earthquakes
started to be felt in other locations.

And this led to the founding of a
seismological laboratory in Bochum,

Germany, all the way back in 1908.
And it also led to the founding of

a seismic monitoring network
in the Silesia Coal Basin in Poland

a dozen years later with the
express purpose of studying these

earthquakes caused by mining.
And so there’s a very long and rich

history of mining-induced earthquakes
and induced earthquakes in general.

So the next historical example
I want to bring up is from this

garden spot in Goose Creek, Texas.

So this is a photo from the 1920s,
and what you see is oil derricks

just everywhere.
And so Goose Creek is on the coast

about an hour away from Houston.
And so this massive production –

so the extraction of oil –
is just going crazy there.

And so this stated earthquakes,
the largest about a magnitude 4.

And, to my knowledge,
these are the only earthquakes

to ever have been felt
in the city of Houston.

So the last historical
example of induced earthquakes

I wanted to bring up
is Lake Mead.

So this is a photo – a recent photo
of Lake Mead. Lake level is low.

But when they started filling Lake Mead
in the early 1930s, you’re adding a

whole lot of weight into this reservoir.
And so the weight of this water actually

started causing earthquakes in the area.
Fortunately, they were relatively small

low-magnitude 3 earthquakes, and
these have more or less waned away.

So we’ve now seen that induced
earthquakes have been occurring for over

different kinds of induced earthquakes.

So what are some characteristics
of induced earthquakes?

When I start thinking about –
when I see earthquakes occurring

in a peculiar place where I don’t think
of natural earthquakes occurring,

the things I start thinking about are,
are these earthquakes close to

some human activity that
could be causing them?

Are they close to a mine?
Are they close to a reservoir?

The second thing I start thinking
about is, are these close in time?

Was there some change
in the human activities

that could be linked
to the earthquakes?

And finally, are these
earthquakes close to the surface?

In general, human activities are at the
surface, or very close to the surface.

So the effects of these activities
are going to be close to the surface.

And so these are sort of criteria
that I start thinking about

when I start thinking about whether
or not an earthquake is induced.

Unfortunately, these are
not hard-and-fast rules.

There are plenty of examples
of induced earthquake sequences

that violate one
of these rules.

And I can actually provide an example
that violates all three of these criteria.

So they’re good things
to think about, but they

aren’t something that
you can directly rely upon.

So why is it that induced
earthquakes are suddenly an issue?

We’ve known about
these for over 100 years.

And what’s really happened is this
dramatic increase in seismicity that I

showed you a little bit earlier in the talk,
is that the earthquake rate has

increased by more than a factor of 10.
You look at this period from 1973 to

about 24 or so earthquakes per year.

And in the subsequent nine years,
we saw nearly three times as many

earthquakes in this
much shorter period of time.

So there’s a
very dramatic change.

And, again, I’m looking at
magnitude 3 and larger earthquakes.

And the reason we’re looking at 3 and
larger earthquakes is that we’re certain

that we’re seeing all of these earthquakes
throughout this study period.

While our recording capabilities and
our earthquake detection capabilities

have changed over time, and in general,
gotten better, even in – back in 1973,

we were able to see all of these
magnitude 3 earthquakes.

So the signal that we’re
seeing here is not a function

of our capabilities but
is actually a real signal.

So let’s take a look at what
the seismicity looks like.

So now we’re looking at a
period of time when there really

wasn’t a lot of induced seismicity.
And you can see that seismicity

is sort of scattered around
the central United States.

You can see that there’s this
structure of seismicity right here.

This is the New Madrid seismic zone.
It’s an area of known natural seismicity.

And you also can see right here
the Eastern Tennessee shear zone.

This, again, is natural seismicity.
But in general, the seismicity is

sort of scattered around everywhere.
You see some small clusters,

but in general, it’s sort of spread
around somewhat randomly.

And the average rate of
earthquakes was about

again, magnitude 3 and larger.

So then, if we look at 2009
through January of this year,

we have 3,200 of these earthquakes.
And if you go back and forth

between these, I would guess
that this figure has more earthquakes

on it if I compare these.
But actually, we just have

so many earthquakes lying on top
of each other here in Oklahoma,

it actually just looks like
there are less earthquakes.

And so our earthquakes over this span
of nine years is now 357 earthquakes,

but if you recall, we had a year
with over 1,000 earthquakes per year.

So this is really actually lowered by
some of the lower-rate earthquake years.

And so this earthquake rate increase
is limited to just a few areas.

So now if we look at –
what we’re looking at here now

is a cumulative count of earthquakes.
And so if we – if we expect that

earthquakes are going to be
happening on a random basis

at more or less the same earthquake rate,
we’d expect there to be a line.

And that’s what we see
right here from 1995 to 2008.

And things start turning upward because
our earthquake rate is increasing.

So what I want to do is really show you
that the increase in the earthquake rate

is just coming from a few places.
Now, I’ve alluded to the fact that

Oklahoma and southern Kansas
is really the source of most of

this earthquake rate increase.
So let’s just subtract out the

earthquakes from Oklahoma
and Kansas and see what happens.

Well, we’ve almost flattened out this
line completely by just subtracting out

Oklahoma and Kansas. And we
know these earthquakes are induced.

But let’s subtract out some
other induced earthquakes.

We can subtract out the
Raton Basin here in

southern Colorado/
northern New Mexico.

We can subtract out the Guy-Greenbrier
sequence in central Arkansas.

We’re flattening out
even further.

And if we take out a few spots in Texas,
we’ve now flattened out completely.

So now we’re looking at a
linear accumulation of seismicity.

So what we’re seeing is that the natural
seismicity rate really hasn’t changed.

We’re just seeing a change
in seismicity in a few areas,

and these are all areas where
we’re seeing induced earthquakes.

Now, the earthquake rate increase that
we’re seeing in Oklahoma is so dramatic,

the earthquake rate in Oklahoma is
now higher than it is here in California.

And so what we’re looking at
here again is a count of magnitude 3

and larger earthquakes.
Shown in blue is California.

Shown in red is Oklahoma.
California’s earthquake rate is

sort of rattling around somewhere
between 300 and 500 magnitude 3’s

per year, with the exception
of these four spikes.

These four spikes correspond to the
timing of large earthquakes in California.

As you would expect with these large
earthquakes, you’re going to have

aftershock sequences, so you’re going to
have a very high earthquake rate.

But you can see it drops back down
the year after all of these earthquakes.

Oklahoma, on the other hand,
is averaging about one or two

earthquakes a year until 2008
when it starts creeping up.

And then, in 2014, it surpassed
California. And for the last three years –

four years – excuse me – it’s been higher
than the earthquake rate in California.

Right now, the projection is,
for 2018, we expect the

earthquake rate to be very similar
in both California and Oklahoma.

Now, I’ve been talking about
magnitude 3’s for a while now.

It’s not just magnitude 3’s,
and it’s actually – we’re starting to

see significant damaging earthquakes.
And so these are two photos from

earthquakes in 2011 that happened
just a couple of months apart that

really brought people’s
attention to this problem.

These were the first really
significant damaging earthquakes

that we’ve seen in
this surge of seismicity.

The picture on the left is from an
earthquake in Trinidad, Colorado.

This is the Raton Basin,
so southern Colorado.

You can see that this is sort of
an old storage facility with

a brick façade, and it lost
a lot of its – a lot of its façade.

And here is a photo from the
Prague, Oklahoma, earthquake.

And you can see, again, this is a
home that lost its entire brick façade.

So we’re starting to
see larger earthquakes.

And even though the earthquake
rate started to decline in 2016,

we started to continue
to see large earthquakes.

And in fact, three of the four
largest earthquakes to occur

in Oklahoma happened in 2016.
The first happened just a few weeks

into 2016. We had the
magnitude 5.1 Fairview earthquake.

You can see some small amount of
damage associated with that earthquake.

Then the magnitude 5.8 Pawnee
earthquake was and is the largest

injection-induced earthquake
that we’ve ever observed.

And then, a few months later,
we observed the magnitude 5.0

Cushing earthquake.
And this earthquake was actually

the most damaging of all of
the earthquakes that we’ve seen

in Oklahoma, despite the fact
that it’s much smaller than

the Pawnee earthquake.
And the reason is, is that this

Cushing earthquake more or less
occurred in town, where these

other earthquakes were really
occurring in very rural areas,

so we weren’t seeing a lot of damage.
And so this is really something to

sort of step back and think about
that we’ve been very fortunate

that most of these earthquakes
are occurring in rather rural places.

And in fact, the city of
Cushing is relatively small –

somewhere around

We’re not seeing earthquakes
in Oklahoma City.

We’re not seeing
earthquakes in Tulsa.

Now, Cushing is also very
important because it’s a location

of very critical infrastructure
in the United States.

Cushing calls itself the Pipeline
Crossroads of the World.

There are nine major pipelines
that cross through there,

including – many of you probably
know the Keystone Pipeline and the

hotly debated Keystone XL Pipeline
go through Cushing, Oklahoma.

And because it’s the
Pipeline Crossroads of the World,

there’s a whole lot
of oil and gas storage.

Approximately 10% of the U.S.
crude oil is stored in this one town.

Now, these storage tanks –
I’ve been out to visit –

they’re about a football field across.
They are gigantic.

And if you look a look at this
Google Earth image, you can see

that there are a whole lot of these.
They are really everywhere.

And this – so in the city of Cushing,
they can store about 60 million barrels

of oil. For the beer drinkers in
the room, that is two kegs of beer.

If you’re not a beer drinker,
think 42 gallons.

So your fridge is going to
be awful full with the milk.

So Cushing, as you can see,
is a very – a important place.

And it has experienced
a lot of earthquakes.

We were fortunate last year not to
see any magnitude 4’s in the area,

but as I mentioned, there was
a magnitude 5 in 2016

and a few magnitude 4’s
in 2014 and 2015 as well.

So the next thing I want to
show you is an animation

of the seismicity in Oklahoma.
I’ve shown you in a graph how these

earthquakes have behaved, but I feel like
this animation really shows it better.

You’ll see things popping up.
Those are going to be earthquakes.

But you’ll also start
hearing the earthquakes.

So the seismicity is really – is going to
sort of start in this area in here.

This is near Oklahoma City.
[popping sounds correspond with dots]

We’ll see the Prague
earthquake happen here.

And then we’ll start to see
seismicity move northward

through the Hunton Formation,
and the Mississippian into Kansas.

So I’ll just let you listen.

[continuing multiple popping
sounds going progressively faster]

[loud popping sounds
happening extremely fast]

[popping stops]
[audience reactions]

- Wow. Good visualization.

- I really can’t put it any better than
that. [laughter] It’s an incredible

earthquake sequence
that we’re seeing there.

It ends in 2016, so really sort of
at the peak of the seismicity,

so we don’t actually
hear the decline,

but you can see what a dramatic
change that we’ve seen there.

Now, you may be saying to me, Justin,
you have all these earthquakes, but I –

you know, Oklahoma didn’t
have earthquakes in the past.

How is it that they’re
having earthquakes?

There’s no faults there.

Well, actually, there are a
whole lot of faults in Oklahoma.

So what we’re looking at is a map of
faults in Oklahoma that was produced

by the Oklahoma Geological Survey.
And they worked in collaboration

with industry – with a whole
lot of industry collaborators

to put this map together.
If you looked at this map

from 10 years ago where
they didn’t have industry input,

there probably would
have been 10 lines on here.

But industry was very generous
and shared a lot of their

geologic knowledge of the area
so that the OGS could get a better

handle on what is
happening in the area.

And so now that you see that there are
faults everywhere, it makes more sense

that you can have earthquakes just
about anywhere in Oklahoma.

And this is – this is really the case
just about anywhere in the world.

There are going to be
faults just about everywhere.

They may not be active,
but there are faults there.

So let’s now walk through the
different oil and gas operations

that can be causing these earthquakes.
Many of you are probably thinking

hydraulic fracturing is the source
of most of these earthquakes.

Well, that’s not the case. We’ll walk
through why that’s not the case.

But this is just a photo of a frack site.
The largest earthquake we’ve seen

associated with frack jobs
is about a magnitude 4.8.

Oil production has actually been
linked to very large earthquakes –

earthquakes as large as
about magnitude 7.

There was a series of three
magnitude 7 or so earthquakes

that occurred in the late 1970s
and early 1980s in Uzbekistan.

Unfortunately, we don’t know
a lot about what was happening there

because there wasn’t a lot
of communication between

the USSR and the
West at this time.

The other two ways earthquakes
are going to be induced is a process

known as wastewater disposal.
This is our real bugaboo.

This is what’s causing the earthquakes
we’re seeing in Oklahoma.

And the largest earthquake that
we’ve seen associated with it is

a magnitude 5.8 – the earthquake
that occurred in Pawnee in 2016.

And enhanced oil recovery –
another injection technique –

has caused earthquakes
as large as magnitude 4.5.

So how is it that these four different
processes cause earthquakes?

Three are injection,
and one is extraction.

So these are the two different ways
that earthquakes can be caused.

And I’ll really focus
on this one here,

and we’ll go through an animation
to look at this second one.

So, more or less, as I said,
we’re either putting something in

or taking something out of the Earth.
And so we’re adding mass or reducing

mass in the Earth. And so we’re going to
be pressing down even harder on faults

or relieving stress on these faults. And so
this can encourage or discourage faults.

And I think, a few years ago, a lot of us
were really discounting this mechanism

as being particularly important,
but a number of modeling studies

that have come out in just the last
couple of years are showing that

this is turning out to be a more and
more important mechanism

for these earthquakes.
But we still – we still think

this is probably the primary mechanism
for most of these earthquakes.

And this is related
to fluid pressure.

So how is it that injection
causes earthquakes?

So what we’re
looking at right now –

here’s an injection well that’s drilled
down into our injection formation.

So we’ll start injecting –
we’ll start injecting fluid,

and the water’s going to go down
this well into our injection formation.

You can see right now this fault is being
clamped closed by the regional stress.

Fluid is going to be flowing outwards
into this formation, and eventually,

this fluid is going to penetrate our fault.
And while it penetrates the fault,

it’s going to be prying that fault open.
And so making it – releasing some of the

stress that’s holding it closed, making it
more likely to slip in an earthquake.

You might want to imagine it
somewhat like an air hockey table.

When the air hockey table is off, your
puck isn’t going to slide very quickly.

But as you turn the air on,
the puck slides very smoothly.

And that’s more or less what we’re
actually seeing here is, as you fill up

the fault, you’re starting to
turn on the air in this fault –

make it easier for this fault
to actually slip in an earthquake.

So now let’s walk through hydraulic
fracturing and wastewater disposal.

These are the two ones
that really think about a lot.

So up here is a photo
from a frack site.

And you can see it’s an
incredibly complex operation.

It’s amazing the amount of
technology that’s going on in these.

This is a photo from
a frack site I visited

a number of years ago
in Weld County, Colorado.

And these are water tanker trucks.
There’s about 100 of these waiting

to inject their water underground
just into one well.

And so these are –
use an incredible amount of water

to frack an individual well.
So it’s a massive operation.

So hydraulic fracturing is
actually a very old process.

It was invented over 70 years ago
in the Hugoton Field.

And, in some sense,
it is intentionally making earthquakes.

You’re trying to make
earthquakes that are very small –

somewhere on the order of a
magnitude minus 2 to a magnitude 1 –

an earthquake that you and I are
never going to feel at the surface.

And it really is a high-pressure injection
intended to increase your permeability.

If you look at this cartoon here,
if you think of a well that, before it’s

been fracked, you might only be able to
access sort of these shaded areas of oil.

And the idea is,
after you fracked the well,

you’re able to access
a much larger area.

Now, a frack job is going to
be typically short-duration.

It’s going to last hours. An entire
well might take a few days to frack.

So it’s a short-duration operation.
And the amount of water that’s going in,

it doesn’t – it sounds like a lot, but in
the grand scheme of things, it isn’t.

It’s about 100,000
barrels per well.

And, after you’ve fracked your well,
your well goes into production.

You extract out your frack fluid,
and you start sucking out the oil.

Because that’s why you fracked
this well in the first place.

So we can look through an
animation here to show you

what hydraulic fracturing is.
So here we’ve started to drill our well,

and this is going to be our
pay formation right here.

And so you’ve turned the well to
stay into this – in this pay formation.

Because you want to extract
as much resource as you can.

So if you can access oil across the length
of this well, you’re going to be able to

get much more.
So right now, the well is bare.

There isn’t anything
holding it closed.

So they pour cement down it.
And they pour the cement down it

so that it has structural integrity
so that it’s not going to just collapse

in on itself due to the
weight of the earth around it.

But now we’re sealed off.
And so what they need to do next

is actually create
perforations so that you’re

able to get the oil
to flow in and out.

And so they’ve lowered a gun and
set off charges and perforated this well.

And so you can see the
holes here, here, here.

And so they pull out the gun
and then start to inject water –

again, at very, very high pressures.
And so the water will flow down

the well – hopefully any second –
and it’ll go out of these perforations.

And as you raise this fluid
pressure higher and higher,

it’ll eventually exceed the strength of
the rock and create these fractures.

And so now you’re able to access
oil that’s in a much larger region.

And so this is just
one frack stage.

In a long well like this,
you’ll frack multiple stages.

And so the next thing we’ll see is that
this area right here will be perforated

and then fracked so that they’re able
to access the oil in this area as well.


So, as I mentioned, wastewater
disposal is the other major source of

these earthquakes. Actually, it is the
major source of these earthquakes.

Hydraulic fracturing really is
only responsible for a very small

percentage of the earthquakes.
But first we have to ask the

question, what is wastewater?
And wastewater, especially

in Oklahoma, is primarily what
is known as co-produced water.

Now, oil and gas are
the decomposed biological

components of
relicked oceans.

And so what this means is, trapped in
the same space as the oil and gas, is the

saltwater that’s left over from these
oceans. And now it’s even saltier.

And so when you pull out oil and gas,
you’re going to be pulling out saltwater.

It’s not really a choice.

And this is just a simple schematic.
Here you can see this is our reservoir

rock here, and you can see the oil
is floating on top of saltwater.

And so, when you pull out oil,
you’re going to get saltwater with it.

It’s not a choice.

And in parts of Oklahoma,
you’re looking at very, very

resource-poor reservoirs. You’re looking
at 20 parts saltwater to 1 part oil.

So you’ve got a lot of water
that you’ve got to get rid of.

And so this is the major component
of wastewater in Oklahoma.

In other parts of the country,
it’s actually spent frack fluid.

It’s water that you’ve
put down to break up the rock.

But in general, operators want to
re-use frack fluids because frack fluid

is primarily freshwater,
and freshwater is not cheap.

So your options to deal
with all of this wastewater is, well,

first, if you can re-use it, that’s
obviously going to be your best option.

The next option is,
well, surface discharge.

Can you clean this water up enough
such that you can use it for agriculture

or put into a river or something like that?
But what is very commonly done is

you dispose of this water at depth
into a wastewater disposal well.

And so a wastewater disposal well,
you’re going to be injecting into a

very porous formation –
something that’ll take this water easily.

And you’re going to operate
this well probably for years

or maybe even decades.
Some of the largest wells,

you can inject over a million barrels a
month – so much bigger than a frack job.

And over the course of history in the
United States over the last few decades,

there have been about 35,000
of these operating in the U.S.

I believe there’s about 30,000
that are active at the moment.

And not actually too many have
been connected to earthquakes.

So this is what a wastewater
disposal well looks like.

You’re injecting down into
this porous formation here.

Hopefully the water goes into
this formation, and you never

hear from it again. And, well, that’s
generally what happens, but not always.

So let’s now think about
what we just learned about

wastewater disposal and
hydraulic fracturing

and figure out which is more
likely to cause earthquakes.

Wastewater disposal is
a long-term process.

You’re operating for
years or even decades.

And you’re
injecting high volumes.

Some of these wells, you’re injecting
over a million barrels a month.

Fracking, you’re injecting over
a few hours or a few days,

and you’re injecting, relatively
speaking, a much smaller volume.

And so fracking is going to be
affecting a much smaller area

over a much shorter period of time.
And so you would expect

that wastewater disposal – to be
more likely to induce earthquakes.

And, well, that’s
exactly what we see.

Wastewater disposal has been associated
with many felt earthquakes and

many damaging earthquakes, the largest
of which was a magnitude 5.8.

Hydraulic fracturing really has been
linked to very few felt earthquakes.

And, to my knowledge, there hasn’t
been any damage associated with

frack-induced earthquakes.
And the largest frack-induced

earthquake is significantly smaller –
a magnitude 4.8.

And, in fact, for some reason,
all the largest frack-induced earthquakes

that we’re seeing are being experienced
in Alberta and British Columbia

as opposed to anything here
in the United States.

And, again, this is despite the fact
there are many, many more frack stages

that have been completed than there
are wastewater disposal wells.

So now I’ve told you that hydraulic
fracturing isn’t a new technology.

I’ve told you wastewater
disposal isn’t a new technology.

So why is it that we’re suddenly
seeing all of these earthquakes?

And what has changed is a
change in drilling technology.

Now, traditionally,
wells were drilled vertically.

And so, again, if this is your
reservoir formation, this is where

you’re going to be
extracting your oil and gas.

And so you’re going to be producing
some amount of produced water.

Let’s imagine it’s
a kiddie pool.

But what has been developed
in the last 15 years or so is they’re

able to steer the wells and
steer within a narrow formation.

And what that means is that they’re
able to produce formations that have

much lower ratios of oil to water.
So you’re able to produce areas

where there’s just going to be
more saltwater coming out.

And the average horizontal well
produces 10 times more water

than these
vertically drilled wells.

And so this is the source
of these earthquakes.

We’re producing way more water,
and this water needs to be injected.

And so it’s this dramatic increase
in fluid that’s generated from

the horizontally drilled wells
that’s the source of this.

So let’s walk through a couple
of historical examples.

And we’ll first
look at how we discovered

that fluid injection
can cause earthquakes.

So this goes back to the early 1960s
to the Rocky Mountain Arsenal.

Rocky Mountain Arsenal
is just outside the Denver area.

And, at the time, they were
producing chemical weapons there.

As you might guess,
chemical weapons have a lot of

nasty byproducts, and you need to
get rid of them one way or another.

And so the Army drilled a deep
well and started injecting away.

Injection peaked at about

And shortly after they started injecting,
they started to see earthquakes.

So here we can see just a count
of the millions of gallons

of fluid injected per month
over the course of 1962 to 1965.

And up here, we’re looking
at the earthquake rate.

And you can see, earthquakes
start up just a month or two

after injection started.
And the amount of earthquakes

correlates reasonably well to
the amount of fluid that’s injected.

They turned off injection for
almost a year in ’63 and ’64,

and you can see the
earthquake rate dropped.

Then they turned the well back on,
and the earthquake rate went back up.

And they abandoned
the well in late 1965

because they realized
what was actually occurring.

A couple important things to take away
from the Rocky Mountain Arsenal.

First, the largest earthquake we saw
associated with it was a magnitude 4.9,

and there was some
damage associated with it.

Here you can see
damage to a bridge.

But another important takeaway
from this is this earthquake –

this largest earthquake happened over
a year after they stopped injection.

And so what this tells us is,
even if you stop injection,

it doesn’t mean that your hazard
has gone away immediately.

So just – this is not a magic
bullet by stopping injection.

The other important takeaway
is in this figure right here.

So here we’re looking
at a cross-section of the Earth.

So we’ve taken a slice into
the Earth, and so this is down.

This is up and down.
And so here’s the surface, and here’s

the well, and this is – this is the bottom
of the well where you’re injecting.

And so now we’re a year and
a half after they stopped injecting.

And so we’ve plotted all the
earthquakes in the area, and you can

see nearly all of the seismicity is

And so this tells us that you can
have earthquakes at very significant

distances away from the source of –
the cause of these earthquakes.

So the next historical study
that I wanted to mention is work

that was actually led here out of
the USGS office in Menlo Park.

Pictured here are
Jack Healy and Barry Raleigh,

and they worked
with John Bredehoeft.

And they managed to convince
Chevron to give them control of

part of one of their fields in
western Colorado in Rangely.

And they wanted to test the
hypothesis whether or not fluid pressure

was important, where I showed you
in that animation what things –

how these earthquakes
are being caused.

And they said, our hypothesis is, if we
inject, we’re going to cause earthquakes.

And then, if we suck out all the water
and return this area to its natural state –

to its natural fluid pressure,
we’re not going to have earthquakes.

And if this is true, this is going to prove
that fluid pressure really is the cause of

these earthquakes. And, as it turned out,
that’s exactly what they saw.

So here, again, in these two figures,
we’re looking at a cross-section

into the Earth.
So these are the wells going down here.

And so we’re looking at a six-month
period here where they’re injecting.

And you can see that there’s seismicity
right at the bottom of these wells,

and then there’s some
seismicity a few miles away.

And then they said, all right,
let’s turn these wells off.

Let’s suck everything back out
and return it to its natural state.

So we’ve now gone from a six-month
period to a one-year period here.

And you can see there
are almost no earthquakes.

And so this really shows us
that it is the fluid pressures

that are causing these earthquakes.
And this is – this is really a remarkable

study, and to my knowledge,
there really has been no other case

where we’ve really been able to
actually sort of control earthquakes.

So I next want to walk you through a
modern study of induced seismicity,

and I – the reason I show this one is
I feel like it was an incredibly thorough

study to try to really make sure that
these earthquakes were induced.

And so, in late 2013,
people in the city of Azle,

which is an exurb to the
west of Dallas-Fort Worth,

started feeling earthquakes –
earthquakes as large as mid-magnitude 3.

And so the USGS,
along with Southern Methodist

University in Dallas, started
studying these earthquakes.

And so originally, these earthquakes
are occurring in Texas.

We don’t have a lot
of earthquakes there,

so we don’t have a lot
of seismometers there.

And if you don’t have a lot of
seismometers, you’re not able to detect

small earthquakes, and you’re not able
to locate those earthquakes well.

And so this is what the distribution
of earthquakes looked like.

And it’s about

So you’re not going to be able to figure
out what well was causing these

earthquakes if your distribution of
earthquakes is 20 miles across.

So we went in and put in some
seismometers in the area.

And we were able to reduce
that cloud to now a 2-mile across.

And so now we’re really able to
understand what is happening there.

And so, with this information,
Matt Hornbach and colleagues,

both here at the
USGS and at SMU,

really considered all the possible
causes for these earthquakes.

He asked the question, could
these earthquakes be natural?

Could these earthquakes be
caused by lake level changes

in Eagle Mountain Lake?
This is a big reservoir right here.

Could it be due to
water table decline?

Texas had been in a very
prolonged drought at that time.

Or could it be related to
oil and gas production –

so extraction of oil – or related
to wastewater disposal?

So they first considered whether or not
lake level changes could be causing it.

And here’s just a plot
showing you the lake level,

and here’s the time
of the earthquakes.

And you can see the lake level is at its
lowest over this period of six years.

But the stresses associated with this
lake level drop are incredibly low.

So here’s the location
of the earthquakes,

and this is showing you the change
in stress associated with this.

And you can see that stress changes are
only located immediately under this lake.

And these stress changes
are incredibly small.

They’re less than a kilopascal.
We generally expect something

to be triggering an earthquake at –
you need an absolute minimum

of 10 kilopascals or more.
And so we just don’t think that this

drop in the lake level could really have
anything to do with these earthquakes.

They also considered
the water table decline,

and the effects were even
smaller than the lake level change.

So they were able to
rule these out pretty quickly.

So the next question was,
could it be oil production?

And if you take a look at
this Google Earth image,

you can see that it’s – it looks like
it’s pock-marked everywhere.

Well, those pock marks are
actually the location of oil wells.

So there are oil wells just
about everywhere here.

And that’s what this
map is also showing.

So each of these symbols
represents an oil well.

And the – and the lines are showing
you the direction that the well is going.

So there are wells
absolutely everywhere.

So it’s certainly possible that these are –
these are related to the earthquakes.

And there’s also two high-volume
disposal wells that we see here and here.

And so what Matt and his colleagues did
is they computed the stresses associated

with both the production of oil and gas
and the injection of these earthquakes.

And you can see that all the earthquakes
are located right at the edge of the areas

of highest pressures associated with the
injection and extraction of these fluids.

And not only that, they were able
to show that the timing of these

earthquakes corresponds very
nicely with the highest stresses.

So we can see right here, these are –
these are the plots of the stresses,

and these stems here are showing you
the timing of the earthquakes.

And at the period of the peak,
the earthquakes are occurring.

So it’s a very clear temporal
and spatial correlation.

So now I want to step away
from a site-based study and

start looking at some broader studies.
So I’m going to be presenting some

work done by Matt Weingarten,
who at the time was a graduate student

at Colorado University and is
now a postdoc at Stanford.

And he and I, and a number of
other scientists, worked on this.

And what he did is he gathered all of
the injection well data for wells across

the central United States, and so a real
herculean effort to gather all this data.

There’s 27,000 wells
that he gathered data on.

And if you look at the well density
in the area, you can see that

most of the wells are
going to be concentrated in

north Texas, Oklahoma,
and parts of Kansas.

And so Matt wanted to ask the question,
are there certain injection behaviors

that make it more likely
to cause earthquakes?

And so, to answer that question,
he said, all right, which wells

could be associated
with earthquakes?

Which well is operational at the time that
there’s an earthquake within 10 miles?

He sort of made the – said that that was
a plausible sort of argument that that

well could be causing those earthquakes.
And so, shown in blue are wells that are

not associated with earthquakes
and shown in yellow are wells that

could be associated with earthquakes.
That doesn’t mean they’re causing

earthquakes, but they
could be causing earthquakes.

And so what he did is he said, all right,
let’s think about the parameters

that seismologists really think
are related to increasing the

probability of causing earthquakes.
And so injection rate and

the total amount that
you’ve injected were the

first things that we
sort of thought about.

And so what he’s done here is he’s
plotted the percent of wells associated

with earthquakes, and we’re plotting it
against the maximum injection rate.

So all the wells that injected

somewhere around 40% of the time
associated with earthquakes.

Whereas, wells injecting over
a million barrels a month or more

were about 80 or 90% of the time.
But obviously, we expect there’s

going to be some amount
of random correlations.

And so if this blue line lies within
these red lines, which is what we would

expect for sort of a random correlations,
we don’t think there’s a real signal.

But as you can see, once we get to about

we’re well above these
lines that indicate that

we’re probably seeing
a random signal.

And so what we can confidently take
away from this is, once you get to these

very high injection rates, you’re much
more likely to be causing earthquakes.

Looking at total volume here,
you really don’t see

a very significant signal,
which we found pretty surprising.

We also didn’t see any correlation
with proximity to basement.

Now, what I mean by “basement”
is the really hard rock.

In general, you’re going to be
injecting into sedimentary formations.

Sedimentary formations
really aren’t strong enough

to actually produce
an earthquake.

Stress doesn’t build up enough
to produce an earthquake,

so you really need these strong
granites to produce earthquakes.

And these are going to be below
the sedimentary section.

So it stands to reason that,
if you’re injecting into the basement,

or close to the basement,
it’s going to be easier for those

fluid pressures to get to those depths.
But we didn’t see any connection

in this – in this study, nor did we see
any connection to injection pressures.

That’s probably actually
related to the quality of the data

as opposed to whether
that connection exists.

And obviously, we’re not going to
be able to control for geologic factors.

We’re not going to be able to control,
is there a fault nearby?

We’re not going to be able to control,
is there a way for the fluids

to get to depth?
So we don’t have all of the information.

And so some work that Matt has
done that is – that is still in review,

but I think is incredibly compelling
is they’ve looked now just at

Oklahoma and Kansas, and so
this is updated through last year.

So there’s about four more years of data.
And so they were able to show,

not only is injection rate important,
but total injected volume is important.

And we think part of the reason is we’ve
now had a number of more years, and

you’re going to expect your injection rate
and your total volume to be correlated.

Because, if you’re injecting
at a high rate for a number of years,

you’re going to have
a lot of total volume.

But they were also able to
show that proximity to basement

is important – so being close
to these hard rocks.

So now we’re looking at
proximity to basement.

So zero is going to be
you’re injecting into basement.

And this is the percent of
the wells that are associated.

So you can see the red line is
showing the percent associated.

The black lines are showing random.
And so you can see, until you get

to about 1/2 a kilometer, so 2,000 feet
or so away from the basement,

you’re more or less looking
at a random association.

But as you get closer and closer to the
basement, or actually into the basement,

the probability that you’re
going to be associated with

earthquakes dramatically rises.
And so we can take away from this

that the proximity to
basement is very important.

Still no evidence
of injection pressure.

And obviously, again,
they can’t control for geologic factors.

So now I’ve walked you
through a number of case studies,

and I’ve shown you some
well controls on earthquakes.

We have – we have an understanding of
what physically is occurring in the area.

So let’s try and do something with this.
Let’s try to understand –

maybe make some forecasts of
what we actually think is happening.

And so the USGS has actually been
doing that for the last three years.

These are – here are images of the last
three reports we issued for 2016, 2017,

and 2018, forecasting the
earthquake hazard for one year.

Now, this is really in the same
tradition that we do earthquake hazard

forecasts for natural earthquakes.
I’m sure you’ve all seen the

hazard forecasts for California.
And those are going to be based

in a 50-year perspective as opposed
to a one-year perspective.

And the reason is,
for these induced earthquakes,

things are changing so dramatically,
and it’s not dependent upon tectonics –

it’s dependent upon,
primarily, economic decisions –

things can
move very quickly.

And these are actually used
pretty widely by regulators,

insurance companies,
and a number of people that operate

a lot of infrastructure like dams, so
that’d be the Army Corps of Engineers,

state departments of transportation, the
Bureau of Reclamation – things like that.

And you can see this is just a map
showing you the areas with

highest chance of damage associated
with these induced earthquakes.

Now, I should point out, an important
thing to note about this is, right now,

we’re just using
statistical methods.

We’re predicting next year’s earthquakes
based on this year’s earthquakes.

We’re not using any physics.
This is what we’re doing so far, but I’ll

show you some work that really is in
progress to try to do a better job at this.

And so we can look at what the
hazard looks like with

induced earthquakes and
without induced earthquakes.

And so, on the right, you can see,
without induced earthquakes, the hazard

in the central United States is pretty low.
But if you include induced earthquakes,

you can see there’s a lot of
hazard here in Oklahoma

and parts of Texas and
the Raton Basin as well.

But, well, you’ll say to me,
Justin, well, come on.

You know, you just told me you
know something about the physics.

You know something.
Why can’t you do better?

Well, we can. We know that
injection is controlling seismicity.

We know how close you
are to basement is important.

And we know the
injection rate is important.

And you can
clearly see this.

We’re just looking at a number of
different regions in Oklahoma.

Shown in blue is the fluid injection rate.
Shown in black is the seismicity.

And you can see that there’s a pretty
reasonable correlation between the two.

We also know
the fault geometry.

We’ve got this beautiful
fault map of Oklahoma.

So why can’t we do better
than a statistical model?

Well, part of the reason is, we don’t
really know the fault geometry.

Now, all of these dots are showing
you locations of earthquakes

that have occurred in Oklahoma.
And all of these lines are showing you

locations of the faults
that were in that fault map.

Almost none of these earthquakes
are happening on those faults.

So we don’t actually know where
these faults are, so we can’t actually

predict the likelihood
of earthquakes happening.

We can predict the likelihood of these
earthquakes happening on these faults,

but all these faults appear
to be more or less dead.

And so we really need to start
thinking about these newly observed

faults where the
earthquakes are occurring.

So we don’t know the fault geometry,
so we can’t really do a real probabilistic

sort of experiment looking at
the faults that we know about.

So we’re going to have to
do something simple.

But we still know a lot here.
We know about the geology.

This is a picture of what the basement
looks like in southern Oklahoma.

It’s been exhumed.

We also have very fine core samples
actually from depth of both the

Arbuckle, which is injection formation,
as well as the basement.

So this where the
earthquakes are occurring.

So we know something
about the geology.

And remember, we know
about the injection history.

We have data on the injection
in all of these locations.

So I apologize now for
showing an equation.

I do my best not to use them myself.

But basically, what we’re looking at the
bottom here are three properties of the

formation. So these are basic
geologic parameters that we know.

And then we take
the injection rate.

So we, again, have all
of that information.

So then we’re able to get our fluid
pressures and the change in the fluid

pressures. And from that, we’re able
to forecast our earthquake rates.

And so let’s
go ahead and do that.

And so what we’re looking at here is the
saltwater injection rate, shown in blue.

And shown in these blue spikes
are the observed seismicity rate.

And shown in red is
our earthquake rate forecast.

I’d say we’re doing
pretty darn good.

We’re matching, more or less,
the onset of seismicity.

We’re predicting things to start ramping
up in 2009, and that’s really when we

started seeing seismicity in Oklahoma.
We matched the peak of the seismicity

in that we’re getting it –
the peak of the seismicity more or less

right when it’s occurring here.
And we also are matching the decline.

So not only are we matching
sort of the general behavior,

I would say that we’re matching
the amplitude relatively well.

So this is really, I think,
doing remarkably well.

I’m using a four-parameter
equation to predict the behavior

of seismicity, and it’s
working really, really well.

And while this is
a very broad perspective,

we can do this on a much
narrower scale as well.

And so here we’ve made forecasts of
seismicity for 2013, ’14, ’15, and ’16.

And the shading is going to be showing
you the predicted earthquake rate.

And the dots are
going to be showing you

where the earthquakes
actually happened.

The obvious takeaways here are,
well, nearly all the earthquakes

are happening in the green areas,
and you’re seeing more earthquakes in

the areas that have the darkest green.

So we’re doing a pretty good job.

So we’ve now developed
a model that we’re going to

start pushing forward to
make these hazard forecasts.

But, well,
there’s a new wrinkle.

And that’s the SCOOP
and the STACK plays.

So most of the oil and gas
development that we’ve been seeing

associated with earthquakes has
been in this area and this area.

And so these are the Hunton Formation
and the Mississippian Formation.

These are
two different plays.

And so you can see there’s a whole
lot of earthquakes in this area.

But development is starting in
this blue outlined area –

the SCOOP and the STACK plays.
And you can see there’s

not a lot of earthquakes here.
Those are the dots.

And in 2017,
you see a few more.

And so this is really representing
a challenge for us.

Because it’s really
hard to forecast seismicity

where you haven’t
seen seismicity.

And there’s not a whole lot of
wastewater disposal in this area.

And what we’re actually realizing is
that most of the earthquakes in this

SCOOP and STACK area are actually
caused by hydraulic fracturing.

So here we’re looking at
a count of earthquakes per week

in the SCOOP and
the STACK area.

And you can see that there’s these
really short bursts of seismicity.

And if you go and look, I’d say, hey,
when are they hydraulically fracturing?

Well, look.
It lines up pretty much perfectly.

So what we’ve learned from this is
hydraulic fracturing-induced seismicity

is more common
than we’ve thought.

And really, the only reason that we’re
able to see it in this SCOOP and the

STACK is because there’s not a lot
of wastewater disposal in the area.

While the areas here in the Mississippian
and the Hunton have a whole lot of

wastewater disposal, there’s very, very
little in the SCOOP and the STACK.

And so you’re able to see the
frack-induced seismicity, whereas,

frack-induced seismicity is going to
be masked by this much larger signal.

In general, based on – based on
these studies by Rob Skoumal,

who’s a postdoc working with us here,
we expect about 1 to 4% of the

seismicity we’re seeing in Oklahoma
is related to hydraulic fracturing,

and the rest is really related to
wastewater disposal.

Now I’ve sat and talked to you here
in California about earthquakes

in Oklahoma. Well, let me bring it
back home for just a minute.

Don’t feel sad. We have our own
induced earthquakes here. [laughter]

In fact, we just got – we’ve got
just about every kind of induced

seismicity you could imagine.
We have reservoir-induced seismicity.

We have wastewater
disposal-induced seismicity.

We have the first hydraulic fracturing-
induced earthquakes ever observed.

So we’ve got just
about every kind.

But it’s a lot more
difficult to identify here.

Just like I was talking about
in Oklahoma where you have

this high seismicity rate
due to wastewater disposal,

here we have a high
natural seismicity rate.

So it’s very hard to disentangle
a natural seismicity rate that’s

very high from a low rate
of seismicity that’s induced.

So let’s wrap up starting to
think about, what is our future?

Where do we go from here?
Well, in Oklahoma,

the earthquake rate is declining.
It’s declined 70% from 2015 to 2017.

And 2018 looks like
we’re going down further.

Great news.

This is – this is following
reductions in fluid injection,

which more or less makes sense.
But this is due to two different factors.

One is economics.
The price of oil dropped from

about $100 a barrel to
about $40 a barrel at the end of 2014.

So for industry, it didn’t make much
sense to be really heavily going after

these areas when there wasn’t
a lot of profit to be made.

There also have been
regulations phased in

by a number of regulators
in both Kansas and Oklahoma.

And we think that has also had
an effect on the seismicity rate.

But there definitely is a question –
the price of oil is starting to increase.

What is
going to happen?

But even though the earthquake rate
is declining, we saw three of the four

largest earthquakes to occur in
Oklahoma occur in 2016

when the seismicity
rate was already dropping.

And this includes the largest historic
earthquake to occur in Oklahoma,

as well as the most damaging
earthquake to occur in Oklahoma.

And still, at the end of the day, yes,
great, the earthquake rate is dropping.

But we had about

magnitude 3 earthquakes
in Oklahoma last year.

That’s still 200 times the
background rate of earthquakes.

So I don’t think the people in Oklahoma
are really happy with that, still.

But it’s not all doom and gloom.
I think these gentlemen,

with John Bredehoeft, really showed us
that we can control induced earthquakes

to some degree by affecting the fluid
pressures that we’re looking at.

And there have been regulatory
successes in a number of different

locations – in Greeley,
Youngstown, and Love County.

But we’ve also learned that
distant earthquakes can’t be

stopped instantaneously.
There’s an area called Paradox Valley

in southern Colorado where we’re
seeing earthquakes over 10 or 15 miles

away from the injection.
So if you suddenly stop the injection,

the earthquakes 10 or 15 miles away
are not going to hear of that for years.

And some sequences also continue
long after injection has stopped.

I showed you, at Rocky Mountain
Arsenal, where we had the largest

earthquake a year
after injection stopped.

We’re seeing earthquakes there

So these can be very
long-lived sequences.

States are taking action.
Seven different states have

enacted regulations to respond
to induced seismicity.

And the EPA, who sort of oversees
all the regulations on injection wells,

has released guidance on
minimizing induced earthquakes.

The USGS – we’re now
issuing hazard forecasts.

Right now, based on these statistical
models, but we’re pushing forward

to take these earthquake rate
forecasts all the way to hazard

with these models
that are based on physics.

We’ve moved past just these
simple statistical models.

And I shouldn’t just toot our horn.
There’s some very nice work

done by colleagues over at
Stanford where they’re also

using physics to forecasts
these induced earthquake rates.

So moving forward, our high
earthquake rates are continuing.

They are declining,
but it appears perhaps that they

can be managed to some degree.
And we haven’t seen any really large

earthquakes yet, and we really can’t
count those out from happening.

But one thing that we need to think
about is earthquakes in the central U.S.

are potentially more dangerous
than earthquakes here.

People are not building to the
same construction standards

in Oklahoma as they are here.
There are no earthquake engineering

regulations in Oklahoma like there are
here, and so buildings are going to

be more susceptible to strong
shaking there than they are here.

Things looking up – our ability to
make forecasts is improving.

We’re finally starting to
integrate some of the physics

that we know is going on to make
these earthquake rate forecasts.

And where there’s been a lot of
incredible collaboration and cooperation

between regulators, between
academic scientists, and industry –

and I know that’s
a little bit remarkable.

I was impressed when
I started working with industry.

But at the end of the day, industry –
this isn’t good for them.

They want to stop the
earthquakes as well.

But I’m a scientist. I think we’ve
made a lot of great progress in this.

But more research is needed.
I think there’s both a great societal

benefit here, in that our research can
really help reduce these earthquakes,

but also, there’s a lot to learn
about earthquakes in general,

both induced and natural,
by studying these earthquakes.

So last, I’ll just leave you with a couple
places you can get more information.

On the left-hand side is a picture
of a handout that I’ve got

in the back that is a plain English
discussion of induced seismicity.

So if you forgot
something that I said,

didn’t get something that I said,
that should cover it.

And we also have a website there
with more information about induced

earthquakes. I’ll end it here and
be happy to answer any questions.


- Thank you, Justin. I’m going to
remind people with questions

to please step up to the microphone
so our online viewers can hear you.


- Hello. This is a technique
question, I guess.

You mentioned a lot of those models
based on points of local stress in

the ground. Can you talk a little
bit about how you measure that?

And, you know, is that
getting easier to measure?

Is that getting [chuckles] harder to –
I don’t know.

What are the instruments used for that?
- Okay. So we actually don’t have

a whole lot of measurements of stress
at depth. That’s very, very difficult.

There are stress sensors that are
buried in boreholes, but in general,

those are incredibly expensive
and not always all that reliable.

Most of – most of what we’re
doing here is really stress that’s being

computed from numerical models.
We’re taking what we know about

the Earth and taking what we know
about the injection and predicting

what the stresses are. Now, there
are a handful of monitoring wells.

And so the USGS, we brought in one.
The Oklahoma Geological Survey,

I believe, has about a dozen,
where they actually have a –

they’ve managed to convince an operator
to let them take over one of their

disposal wells. And they’ve basically
dropped a float into that well.

And, as that float goes up and down,
we know what the fluid pressure is.

And so that is incredibly valuable
information because otherwise we have

no way of knowing what the
fluid pressures are in this region.

And so we’re learning a lot
about what’s happening.

The data is slowly starting to come out,
but I think it’s – I think it’s really

exciting, and I think we’re
going to learn a lot more from it.

- I have heard that there are some
climate change deniers. [laughter]

And I’m wondering
if there are also

induced earthquake deniers, or if
this is generally accepted science.

- Well, I think, within the scientific
community, it’s not under debate.

Certainly, I’ve run into people that
want to sit and argue with me.

But I think the number of those people
really is declining over the years.

I think – you know, I’ve been
going to scientific conferences

on this for seven years now.
And six, seven years ago,

industry wouldn’t deny that
Rocky Mountain Arsenal was induced,

but they denied that oil-and-gas-induced
earthquakes existed.

And a few years later, they’re,
like, well, maybe. [laughter]

And now they’ll admit that there
are earthquakes that are induced.

They won’t admit that they’re
inducing earthquakes. [laughter]

Actually, a few operators are
now starting to admit that.

So there’s a progression here.
And, like I said, I’ve actually been

really impressed, in general,
by industry and that,

at the end of the day,
they want to fix the problem too.

So it really hasn’t been a hostile working
relationship. In general, it’s collaborative.

- Continuing on that, do you –
have you ever faced any resistance

from companies in terms of
the research you did at all?

Especially the Rangely experiment.
That sounded like it was

especially collaborative.
So, you know, are there instances

that you think – you can think of
where they weren’t collaborative?

And question two.
It may sound like a stupid question,

but when you extract oil out of the well,
and you said there’s saltwater

coming out as well, can you not just
put that saltwater back after you’re

done with the extraction?
- Okay. Both good questions.

So, like I said, in general,
our relationship with industry

has been
pretty collaborative.

Sometimes they give us data.
Sometimes they don’t.

And, in fact,
a lot of the time, they don’t.

And that, in general, has to do with their
lawyers as opposed to their scientists.

And I primarily work with their
scientists, as you would expect.

Certainly there are scientists from some
oil and gas companies that walk straight

past me at meetings. [laughter]
And that just sort of is what it is.

But I think I’ve had maybe one
hostile interaction ever with an oil

and gas representative, and he and
I had known each other for years.

And I – it was – basically there was a
press release that his bosses didn’t know

about, and they came after him, and, you
know, the next day, everything was fine.

So in general, it’s been really positive.
With Rangely, that was remarkable.

That was 50 years ago.
And I don’t know if Barry, Jack,

and John are – were just
incredibly charismatic men

that convinced Chevron to
give them control of their field.

Or I think business was
done a little bit differently.

People weren’t as concerned
about liability at the time.

Your second question.
Just remind me again. I’m sorry.

- Saltwater.
- Can you put the saltwater …

- Oh, can you – can you re-inject
the saltwater. That’s a great question.

And so that is actually more or less
what’s called enhanced oil recovery,

where you put water
down into the same formation.

You can do that, but in general,
it’s not being done in these locations.

And the reason it’s not being
done in Oklahoma is they’re

extracting these out of what are
known as generally tight formations.

These are formations
that aren’t very porous

that aren’t going to
accept fluid very readily.

And also, you have this problem
that you have very little oil

and lots and lots of water.
So if you’re putting more water in there,

you’re going to be, in some sense,
diluting your supply.

And so you really want to
get the saltwater out.

That obviously would be
a preferable solution.

But, as I showed at the beginning
of the talk, enhanced oil recovery

can cause earthquakes, but in general,
we wouldn’t expect it to because

there’s going to be a fluid-mass
balance and a fluid-pressure balance.

But it doesn’t totally work in this
location. It just doesn’t work.

- I’ve got two questions as well.
First, about four or five slides back,

there’s one where you showed
a temporal chart at the bottom

of the wastewater injection
correlated with seismic activity.

And you overlaid it on and showed that
there was a one-to-one-correlation.

- Hmm, yeah.
- My question is …

- So like this?
- No. A little bit further.

There were individual stripes of red.
- Oh, hydraulic fracturing. Sure.

- Oh, there you go. Yeah.
- Mm-hmm.

- The implication is, is that each one
of these things is very punctuated,

where there’s almost no activity,
and then a whole lot.

- That’s right.
- That seems strange.

Can you explain that a little bit?
- It’s a great question.

So we think what – because hydraulic
fracturing is such a short-term process,

and it’s a very high-pressure blast,
but it’s a low amount of water that’s

going in. So we expect the pressure
blast to drop of really dramatically.

And so that’s really
what’s happening here.

You’re not affecting a very large region.
You’ve just blasted it really hard.

You have earthquakes
during that quick blast,

and then they – it more or less goes
away. In general, most frack-induced

earthquakes are within two or three
days of the – of the frack job.

I’ve seen earthquakes 30 days after,
but that’s incredibly rare.

- Good.
Okay, second question.

On the map of the U.S., you mentioned
that the SCOOP and STACK area and

the places in Oklahoma have a very
low background of natural seismicity.

- Mm-hmm. Mm-hmm.
- The place, however, which does not

have low background is the
New Madrid Fault in southern Illinois.

- That’s right.
- The injection and – that sort of

adds a joker to the
background of this natural stuff.

And that, frankly, scares the
bejesus out of me after what happened

in the 1800s there.
- There isn’t a whole lot of

oil and gas exploration right
on the New Madrid seismic zone.

There is some in southern
Illinois and southern Indiana.

But we’re not seeing any seismicity
associated with it, but – yeah.

- So far.
- So far. So far, yeah.

- This question is partially political,
so you may have difficulties with it.

[laughter] But because of the fact that
the liability could gradually build up,

and therefore there could be
a lot of lawsuits against the

fracking companies by – when you saw
the earthquake damage of some houses,

first thing you do is get
a lawyer and sue the company.

Have you
seen that happening?

And if so, has the Trump administration
starting to clamp down your research?

- There are – there are ongoing
lawsuits in a number of

different locations where oil and
gas companies are being sued,

either in class-action suits
or by individual people.

To my knowledge,
none of these have been resolved.

So far, nobody upstairs has been
complaining about my research.

- Good.
- And I hope it stays that way. [laughter]

- So I think I heard you talk
about pressure and stress.

I don’t remember hearing
much about energy.

So if you compute the energy you put in,
does it equal the earthquake energy?

- No, no, no.
The amount of energy – basically,

so the amount of mass times the
change in elevation is much, much lower

than the amount of energy
that’s released in these earthquakes –

many orders of magnitude.

- So it’s just the intrinsic stress in the 
[inaudible] that you release?

- It’s stress within the Earth
that’s being released, yeah.

- Okay. And the other question is,
if you can induce earthquake by

injecting water, I heard that tides don’t
create earthquakes, so why is that?

Just the magnitude of
the pressure involved?

- Yeah. So people have been looking for
tides causing earthquakes for decades.

Actually, there’s some studies
going back to the 1800s.

And people see connections
in the most extreme cases

where you have these massive tides.
But in general, you don’t.

And there’s – a big part of it is just
the pressure changes are so low.

And that’s the big source of
why we’re not seeing it.

- Okay, thanks.

- Hi. First, thank you for your
presentation. It was very good.

Speaking about delayed and spatially
removed events, you know,

after injection stops, or far away
from injection, I’m just curious what

the travel time through the rocks are.
I’m sure it varies with all kinds of stuff,

but I’m sure there’s fluid transport.
There’s also pressure waves which,

you know, wouldn’t necessarily
mean the fluid’s moving.

So I’m just curious, can you just
put some numbers on how fast these

things are moving through the rock …
- [laughs]

- … at all? Or not really?
- It’s pretty hard.

It’s primarily going to depend –
be dependent on your permeability.

So that’s a characteristic of your rock.
And, you know, there’s going to be

sort of a general permeability
of the rock, but also,

these rocks are
going to be fractured.

And so the fluid pressure can transfer
much more quickly in the fractures.

And so, if you have a long fracture,
it’s going to travel much more quickly.

Yeah, it’s – and an important thing to
remember is that it’s not that the

water actually has to get there. It’s just
that the fluid pressure has to get there.

So if you think about, like,
your hydraulic brakes.

It’s not like the brake fluid that
you’re stepping directly on is going

all the way to your brakes. It’s just
that pressure that’s being transferred.

- Right, but is there a – there’s still
a pressure transmission rate, right?

- Yeah. Well, so it’s – it would be
hard for me to put a number

on how quickly it’s going to move,
but relatively quickly.

I mean, you can move a kilometer
in a month, no problem.

But you also need to consider – if you
recall, I talked about two mechanisms.

I said fluid pressure
is the bigger source of it,

but also those solid
stresses are important.

And so the recent studies that have
been looking at poroelastic stresses –

so solid stresses – those stress changes
are almost instantly transferred.

And so it’s a tough
game to try to figure out.

I mean, that’s – and that’s one way
we’re able to figure out that

poroelastic stresses are important is these
stresses are getting there too fast for it

to be a fluid pressure-based thing.
- Thanks.

- We live in Monterey County,
where fracking and other

oil-related issues are very hot,
including – we’ve got an election

where Chevron is pumping
hundreds of thousands of dollars

against a very good
anti-fracking candidate.

What can you tell us about our area –
about – you know, what could –

what we should be
watching for, things like that?

- Ah, that’s a – it’s a good question.
So, you know, the Monterey Shale

has been known about for,
you know, probably a century or so.

And a number of years ago, people
were really going – were saying it was

the next Oklahoma or the next Texas.
And we’ve more or less figured out

that that resource is much
smaller than we originally thought.

I think it’s about 4% of the size
of what was originally thought.

So what can be done there is
what can be done in Oklahoma.

It’s just enhanced
earthquake monitoring.

Obviously, we’ve already got
a good monitoring network,

and so looking for connections
between the timing of frack jobs

and the timing of earthquakes is
sort of the obvious first step.

- Thank you.

- Yeah, two questions.
A quick one is, what sort of pressures

are you talking about when you’re
pumping this stuff back into the ground?

And then the second one, though,
is more – you haven’t mentioned

anything about offshore and
if there’s any marine oil deposits

and stuff, that this comes
into anything in those areas.

- Okay. So – so, sorry,
the first question was …

- Just how much pressure …
- How much pressure.

So, actually, a lot of these wells, they’re
actually not injecting under pressure.

They’re pouring the
water down the well.

And so it’s literally just the weight of
the water that’s increasing the pressure.

So you’re looking – in some of the
lower areas, you’re looking at

tens of kilopascals,
but with the monitoring –

the pore pressure monitoring that
we have, we’ve seen a rise of nearly –

I believe a megapascal over the course
of a year. So a lot. It can be very high.

As far as offshore, truthfully, I’m not
particularly knowledgeable about

what is really happening out there.
I mean, those are – in general,

what’s happening offshore
are older reservoirs,

so areas that have been
exploited for a long time.

And so those are going to be areas
that are going to be high-cut.

So the areas where there’s a lot of
oil and not a lot of saltwater.

So if that is the case continuing forward,
I wouldn’t expect there to be

a high probability of earthquakes,
but it’s really going to be dependent

on what the activities are now in …
- But there’s no – there’s no actually

pumping and recharging
back in offshore [inaudible] …

- I don’t – I don’t know what
the operations are offshore.

- Hi. I have two questions.
One on behalf of my wife.

Who was too shy to come up
here and ask it. [laughter]

The water that’s being injected,
is it clear, clean water?

Or does it
contain chemicals?

And does it have any
effect on drinking water?

- In general, the water is pretty nasty.
And that’s why they have to

inject it at depth. And now,
the water is nasty, not because of

any oil and gas operations.
It’s because it’s nasty to begin with.

It’s usually 100,000 parts
per million dissolved solids.

So that’s pretty much salt. So that’s
three or four times saltier than the ocean.

So it’s …
- Wow.

- It’s really bad stuff.
And there’s also going to be things

that are going to be leached out.
There’s often arsenic and things

like that in there, and so …
- Geez.

- But, in some areas, the water that
comes up is actually very clean.

And either a very minimal amount
of effort needs to be cleaned –

needs to be taken to clean it,
or even sometimes no effort.

But in Oklahoma,
it’s generally incredibly salty.

- But does it have any
effect on drinking water?

- No. So they’re injecting these
into very – in general, very deep wells

far below the drinking water aquifer.
- Okay.

And my other question –
I wanted to take you back to

one of your earliest slides.
You don’t have to go there,

but it was the picture of the
Joker and threatening to …

- [laughs]
- I think – was it dumping seawater

into the San Andreas Fault?
Was that it, or …

- Yeah. Yeah. It’s Lex Luthor trying
to put water into the San Andreas …

- San Andreas Fault.
- Yes.

- And it just put in my mind, you know,
we’re constantly hearing in the Bay Area

about the threat of the Big One
coming and, you know,

all of these percentages and so forth.
And we all sort of expect it eventually.

And earthquakes, as I understand them,
are due to the relief of stress,

and the stress is constantly building up.
Could there be any beneficial application

of this to reduce the stress
so that we had a whole bunch

of smaller earthquakes
rather than the Big One?

- If only that was the case.

Well, there’s a couple
of different problems.

One is, is that – well, like we’ve
seen in Oklahoma, they can’t control

how big these earthquakes are.
And so we wouldn’t be able to

control how big the earthquakes
are in Oklahoma, and so –

or, here in California.
So, you know, what’s going to

happen is – Earth is under control.
We’re not in control.

And the other problem –
it’s an energy problem.

So each magnitude that you increase –
each increment of magnitude

you increase, the amount of
energy increases by 30 times.

So to release the amount of energy
that’s released in a 7,

you need 30 magnitude 6’s.
Or, give or take, 1,000 magnitude 5’s.

Or let’s say we want 3’s. Those probably
aren’t going to cause damage.

You’d need a million magnitude 3’s.
And, at that point, I might just

take the magnitude 7 and move on
with my life. [laughter]

So it just doesn’t work from an
energy budget perspective, either.

But, yeah, if I was able to do that,
I’d be a rich man. [laughter]

- That town where are
the pipelines converge?

- Cushing.
- Cushing.

I would hope people
are looking at that?

- Oh, of course. Yeah. So we’ve got
a lot of monitoring in the area.

The operators in the area
take this very seriously.

They’re running their own seismic
networks to monitor what’s happening,

and they have very specific
procedures in how to respond

if there’s a certain
amount of shaking.

And it’s generally very low, and so
they have to inspect every single tank.

And obviously, they’ve got a lot
of procedures in case – to respond

in case of a failure of one of these.
Because it would be pretty catastrophic

if one of them failed, much less
a number of them failed.

- One more quick question.
This is about [inaudible] in wastewater.

I know that, during the operation
of wells, there’s a whole lot of

wastewater that can come out for
the whole life of the well, basically.

And it’s a problem on land, where you
don’t want to put it in the sewer system.

My question is about offshore.
Is it regulated in United States waters

where they just can’t dump that
stuff straight back in the ocean?

And same question about
overseas – foreign places.

- I’m going to say I don’t know
to both questions. [chuckles]

As far as –
as far as injection on land,

that’s controlled by
the Safe Drinking Water Act.

And so that’s where the
regulatory authority comes from.

I’m sure there’s some amount
of regulations as far as offshore.

Internationally, what is done,
it’s really highly variable.

I’ve worked with colleagues in Italy and
colleagues in Colombia on similar issues.

And everybody has their
own sets of regulations.

But, yeah, it’s really highly variable.
And, in general, the regulations

for this sort of thing
weren’t made for earthquakes.

They were made for keeping water clean.
And so people are just sort of

figuring out how to deal
with it as it – as it comes.

So I don’t think anybody –
really any country, at least,

has a real organized way that
they’re responding to things.


- Going back to your
equation that you showed,

can you tell us how the stress,
or the pressure, induced earthquake

related to the magnitude
of the earthquake?

And also, in Middle East, they are –
they are doing all these oil productions.

Have they seen any kind of – these
kind of induced earthquakes as well?

- So are you talking about this
figure right here? Is that …

- The equation.
- The equation.

- Oh, the equation.
Oh, so no, we’re not forecasting the

magnitude of the these earthquakes.
Basically, we’re forecasting

an earthquake rate.
There are sort of traditional ways

that we’re able to take sort of
a total earthquake rate and more or less

back out the number of
magnitude 4’s that we’d expect

and back out the number of
magnitude 5’s that we’d expect.

And so those are the sorts of
relations that we would use.

As far as what’s happening
in the Middle East, I don’t know

a ton about
what’s happening there.

So I’d be loath to really
comment at length about it.

I do know there are induced earthquakes
there, but I don’t know much more.

- Thank you.

- Okay.
Any more questions for Justin?

Nope. Okay, well, I want to
please remind you to –

do come back for our July 26th
lecture on acid mine drainage.

And please give Justin one last big
round of applause. [inaudible]


[Background conversations]