Welcome to this talk. My name is Zukman.

I'm in SAS and at the water institute at

ASQ. I've been here for about a year and

I I'm happy to discuss this idea that

we've been floating around for a bit

now.

And uh in terms of the talk style and so

on, if you want to stop me as I'm

talking, that's totally fine. Basically

the the

idea here is that we all face the

impacts of climate extremes around the

world. It's a global problem. I'm not

interested in climate change per se. I'm

interested in climate extremes and what

to do about it and the idea that we are

putting forward.

No.

>> Oh uh

>> yeah. Okay. Got it.

>> Okay.

>> Yeah. So we are calling it weather

jujitsu which is the idea of can we use

the power of the atmosphere to control

the atmosphere rather than trying to

make huge changes in the game. So we

it'll be interesting to see what you

guys think of that as we go through. So

what I'm going to do in the talk is

we'll talk a little bit about climate

extremes, give you some examples and

what I'm working towards is the idea

that a large variety of mid- latatitude

climate extremes can be connected to

basically exactly the same mechanism. So

that sets up the idea that if you can

mess with that mechanism, we have

multiple gains to be achieved associated

with that. So that's what we are setting

up. I talk a little bit about the role

of geoengineering or decarbonization and

the main point I want to make there is

even if we succeed there climate

extremes and their impacts are not going

away. We still have to deal with that

and that's the motivation for

in short this is this saying is

attributed to Mark Twain. Everybody

talks about the weather but nobody does

anything about it. Turns out actually

it's not due to Mont Twain but one of

his buddies Charles Dudley Warner who

came up with that and Mont Twain

appropriated it and of course got famous

with that. So the thought is maybe we

ought to be thinking about that as our

plan for the 21st century is how do we

actually monitor the weather. Um and the

argument I make in short is this.

Historically we have spent billions of

dollars on building infrastructure to

protect us from climate extremes.

We didn't have so many people then. We

now have a lot more people, a lot more

exposed and most of the stuff that we

built in the previous century is now

aging and needs to be replaced and the

replacement cost is in the hundreds of

trillions of dollars which is not going

to happen. So if you're thinking about

spending that kind of money at all, we

should be looking at alternative. So the

last part of this is like okay how are

we going to do this and we take the

inspiration from the idea of chaos

theory or sensitivity to initial

conditions uh which traditionally has

been presented as hey the weather is not

predictable because the equations that

govern it are nonlinear small querations

amplify and so you're lost. So the idea

here centrally is okay we can't predict

so we are screwed on that front. However

if small perturvations amplify maybe we

can introduce small perturvations

deliberately and then see if they can

amplify in directions we want them to

go. So that's the idea and so we what

I'll show you is that we are able to

demonstrate and others have demonstrated

this that in idealized models we can do

this very well. Um the question is how

to make it happen for the real jet

stream dynamics and we'll just be

speculating on that at the present time.

So getting started this is a time series

which is kind of interesting uh that's

put out

by climate.gov in this particular case

and it illustrates losses associated

with billion dollar events. So each

event in question here ended up with a

loss of at least a billion dollars is

the idea. So that's the y-axis on the

left hand side. Uh the actual total cost

is on the right hand side. And you can

see that this has been growing

dramatically. But more importantly,

if you look at the colors, what you see

is that the dominant factors are storms,

tropical cyclones, and floods. In that

order, droughts are talked about a lot.

Droughts are this little orange thing at

the bottom. So what you see with

droughts is that our drought losses in

the United States um are relatively

constant every year. We always have a

drought somewhere or the other. It's not

something that is spectacular, but the

other things have been increasing. So

people view this as a climate change

problem and we'll be talking about that

in a second.

Second thing here is that this

observation uh and this is from

group that is putting together

information from the reinsurance

companies and they point out that

everyone focuses on the big event but

what's been happening is that the

cumulative losses associated with

smaller events that people don't talk

about nearly as much as what's been

going on. So we will discuss that a

little bit further and that's here. This

is from Roger PL Jr.

Somebody wants okay to care of it. This

is from Roger Pely Jr. who is the son of

Roger PL senior which makes sense. Roger

PL senior is a card carrying

meteorologist from Colorado State

University. Roger PLK Jr. is a social

scientist who's a climate change denier

which is an interesting domestic

situation perhaps. But what he presents

here is a typical picture that he has

presented again and again every few

years. And the picture is he takes those

losses that were in the previous

picture. He divides them by the GDP and

then he shows that they're modestly

declining.

Okay. So his view is that okay the

reason these losses are going up is

because we have more exposure. We have

more property value. We have more people

through climate change that doesn't

exist. Okay. My reaction is I don't

really care to join that argument. I

have a personal view that one of these

people is inherently stupid. But we

won't get into that. My point is this.

If we think about what this guy is

really showing, he's saying even if you

manage carbon better, even if you solve

the carbon problem, as GDP grows, which

we expect it will, and as the

populations grow, which we expect they

will, if the percentage of GDP is

constant in terms of losses, which is

what he's sort of arguing, then we are

still going to have those losses. Those

are not going away even after

decarbonization. So we need to address

that particular problem and what's our

strategy for doing that.

Okay. So to kind of get started on

thinking about that, the first lineup

that I have here is we have 92,000 dams

in the US. The median age of these dams

is now 67 years. 5,000 of these are

owned by the federal government. There

is a budget behind it and we think they

are being maintained. Although some

people question that the what about the

other 87,000? These are owned by

municipalities, owned by states, owned

by private companies. And to give you

just one example, there are about a

100,000 of these dams in the state of

Pennsylvania. And there are three dam

safety inspectors.

Data on how much money is being invested

in keeping these afloat non-existent.

Okay. So that's a concern. So now here

what you see is that since the year 2000

up to 2023 which is you know when we

finished this work and published the

paper 552 dams had failed in the US. So

that's about one every two weeks. Now

arguably these have not been large dams

but that's part of the story. There's a

log scale as to number of dams versus

size of dams. As the size of dams goes

up the number of dams goes down. So the

question is, you know, when you talk to

the dam safety organizations and to the

engineering community, we design these

dams for the thousand-year event or the

10,000year event from failure. So one

failing every two weeks is not a good

story. What's the return period actually

associated with the rainfall that led to

the failure of these dams? So that's

what we investigated. And you have a

bunch of different plots here. Just

focus on the bar graphs for now. And a

means that how big was the rainfall

event associated with the failure event

on the day of the failure or the the

largest day preceding the failure of

failure event. Okay, so that's a K5

means that you look at the five days

preceding the maximum rainfall and how

much that rainfall was. K30 is the 30

days preceding. So if you look at those

what we see is that those return periods

that's 100 these are around five or 10

years very very far from the 10,000 year

design that people are talking about. So

what people are talk about now is that

under climate change how will the

probable maximum precipitation or the

10,000year event change and what we are

finding is that things are failing at

very modest levels. So there's no point

talking about the 10,000 year event when

you have failures at this time. Story

changes a bit when you look at J5 and

J30. What that refers to is what's the

probability of exceedence of both of

those things happening. The rainfall on

the largest day before the event of

failure and the five days preceding that

or the largest event and the 30 days

before that. So those are starting to

creep up close to the 100redyear event.

still nowhere close to what we are

talking about as failure. The most of

the climate change attribution

literature is focused on has the one day

rainfall event gone up has the 1 hour

rainfall event gone up. What we are

seeing here is the joint event of

wetness followed by a wet event that's

leading to the

similarly.

Okay. So we covered that.

This is data from the national flood

insurance program which is by the way

insolvent at the moment. Okay. The

national flood insurance program in the

US is designed so that everybody who's

in a 100red-year flood plane has to buy

insurance. No choice. Okay. So that's

the idea.

There are three frames here that we look

at. The first one is insurance claims.

These are at the household level

aggregated to the county level. The

second one is presidential disaster aid

declarations. And the third one is money

put up by the federal government or a

state government to buy out people's

property who are in locations which is

which are perpetually getting flooded.

So they want to pay for it any okay. The

bar graphs show you the return periods

associated with these claims. Remember

the design is a 100redyear flood plane.

The median return period associated with

these rural, suburban and urban under

three colors is around five years or

less. It's not the 100y year event we

are worried about. We are having a lot

of payouts of these things and that's

leading to the insolvency of the

program. There's more on this story but

mostly what I'm trying to create the

picture for you is that it's really

obviously a major hurricane is a big

threat. We have to worry about that. But

we are actually getting significant

losses associated with relatively modest

events. If you go and examine the nature

of those modest events, these are

fronts. These are frontal mechanisms

that are bringing growth kind of storms.

Persistent

input of moisture into the area followed

by an event that crosses a threshold of

some sort.

>> It's also the case that um

urban settings are designed for very low

return periods. Yeah. Um so I mean we're

used to designing dams for extreme

events but urban drainage is

>> 2 years or 5 years.

>> Yes. Yeah.

>> So if you see the urban guy is at five

>> the rural guy is at two. So it's

happening every year to some of these

people. It's crazy basically

and you know as you would expect

presidential disaster de declarations

have a larger return period but still

it's not anything that you not thinking

about. Okay. So our def factor

adaptation strategy this is what I

wanted to summarize with. We have aging

dams that are failing. We just need a

major dam to fail and people will wake

up on that and that luck has not

happened. Although we came closed in

2017.

Um

we have a flood insurance program that

faces insolveny at this moment and all

attempts to change that transfer the

risk onto the people that they are

currently protecting. So people are not

happy about that. So that's kind of our

strategy.

Um, moving to other types of events from

floods. U, this is the great Texas

freeze February February 11 to 20, 2021.

Um, the damage associated with this

event was 200 to300 billion. It exceeded

the largest damage from a hurricane that

we have experienced. That's to give you

a sense of the scale associated with it.

Electricity went out that that affected

10 million people. Water pipes burst

everywhere. So there was no drinking

water supply and 200 people died. 3.8

million fish. So there's an ecological

dimension to it and you know

transportation impacts. It was presented

as unprecedented and an example of what

climate change is doing to us. So the

paper we published actually looked at

this is 2021 the evented question 1 day

3-day and 5day temperature profiles

associated across the country. So you

can see this deep freeze that happens

over the midsection of the country and

is represented at all three time scales.

But if you look up we have 2011 1989

1983 1951 all of them with very similar

events. There was no excuse for a lack

of preparation.

It could be climate change, but it's

hard to make that object. Okay. What

causes this? I'll be dwelling on that in

the next few slides. It's caused by a

very deep trough that basically comes

down from Canada into this area and

brings cold air down and persists. Okay.

Part of the jetream dynamics that I'll

be building up. Uh and you saw this

happened at various other events. If you

don't care about Texas, but you care

about Kansas, if you care about

Missouri, you'll see that this is a very

routine event in those places.

So, what is the jetream? Basically, most

of you probably know, but just for

completeness, we will develop that idea.

The thermodynamic engine for the planet

is basically at the equator. That that's

where you get most of the heating. So

you end up with convection and quite a

bit of that heat needs to now be

dissipated to areas which have lower

temperatures. So if I if I look in this

direction basically I'm looking going

from equator to pole because the pole is

the other end point of this particular

engine because it's until it all melts

it's a constant temperature location. So

that's basically what defines this

particular heat engine. And what you end

up with is a convection cell here that's

called the adi cell, a synchronous cell

here called the fereral cell and a polar

cell. And looking in this direction,

this is the location corresponding to

the subtropical chair. That's the

location corresponding to the polar

chair. And

the features associated with this are

what I was developing. This is our top

view looking from the pole as to how the

winds are organized. And I can talk a

bit more about it, but there's a nice

little video that I will show you. The

basic thing that happened for Texas was

that if the polar vortex is stable, the

polar layer is contained within the

polar jet stream as it's shown on the

left. But if it is disrupted such that

you end up with this big wave in the

polar jetream, you end up with a cold

warm cold warm high low pressure

sequence associated with it. And that's

essentially what was the story

associated with that particular event.

Similarly, if I look at heat waves,

yeah, this is the UK and this is from

the UK office. They were trying to

explain why the last week lay there was

hot and what you see there is the

curvature of the polar jet stream is way

to the north. So you end up with that

situation. And if I expand that further,

you get synchronous heat waves or

freezes across the whole planet

depending on what's called the wave

number associated with the jetream,

which is how many waves around the

world, how many high sequences you get

around the world. So if we start

thinking about these extremes, then what

we are going to end up thinking about is

the manner in which this particular

system basically works.

So here's uh some dynamics associated

with it. Things respond to boundary

conditions and to initial conditions.

That's how in physics we have started to

learn about how these systems work. So

in this particular context, what's being

shown is what happens to the jetream

dynamics during Elino conditions in the

tropical Pacific versus the opposite

side of that which is liner conditions.

And essentially what that does is it

changes the equator to pole temperature

gradient. It changes the land ocean

temperature contrast. And then those two

features combined to determine where the

mean position of the subtropical jet and

the winess of it is manifest. So that's

one example. As we do global warming, we

end up with a weaker equator to pole

temperature gradient because the poles

warm much more than the equator. So that

leads to a condition which is almost

like a perpetual summer, a weaker

jetream and a shift in the jetream

location. So boundary conditions are

important from that point of view. But

I'm going to focus on the weather time

scale rather than the climatic time

scale. So I wanted to show you this.

Back in 2010, Russia experienced one of

its most severe wildfires and heat

waves. While Pakistan witnessed its

super floods, of which the 2022 flash

floods are a cruel reminder. While these

two events were over 2,000 km apart,

they were connected to a single

meteorological event. The Rosby waves

named after KL Gustav Arvid Rosby who

first identified them. While these

events were different in nature, the

waves caused a high pressure pattern

over Russia while influencing downstream

wind patterns in the Indian subcontinent

leading to the floods. But what are

rosby waves and why have they gained

relevance over the last decade? Rosby or

planetary waves occur within the earth's

ocean and atmosphere. They move

continuously along with the rotation of

the planet. These can be classified into

two categories, oceanic and atmospheric

roby waves. Unlike surface tidal waves

which break as soon as they touch land,

oceanic roby waves are found along the

thermocline or the layer where the warm

surface waters mix with the cool deep

waters. They move about in slow

undulating waves for hundreds of

kilometers in the westward direction. It

can take them a decade or more to cover

the surface area of the ocean. Their

interaction with phenomena like the El

Nino influences climate across the

world, sometimes causing high tides and

coastal flooding. Atmospheric Rosby

waves are high altitude winds in the mid

latitudes. When the Arctic jetream

becomes more wavy and flows into lower

latitudes, that meandering is referred

to as Rosby waves. When they swing up,

they transfer heat from the tropics to

the poles. And when they swing down,

they carry cold air towards the tropics

to maintain some balance in the

atmosphere. Any anomaly in Rosby waves

can cause disasters in seemingly

disconnected parts of the earth.

Normally, these waves move eastward,

which implies that weather systems move

with them. But if this eastward movement

slows down or freezes, then high and low

pressure areas persist over specific

regions like they did in Russia back in

2010. While the wildfires and super

floods did not happen together, their

impacts were felt around the same time.

Rosby waves are composed of two

intricately linked types, synoptic and

forced waves. Synoptic waves move

quickly only formed by the atmosphere,

but they also contain minor static or

forced components. Forced waves are

formed by interruptions of mountains and

temperature differences across

continents and oceans. These contain

minor synoptic components as well. These

two types interact with each other

within an atmospheric channel called a

wave guide. This interaction allows

their strength to increase and influence

the weather. This interaction is also

called wave resonance. While we do not

have extensive historical data on Rosby

waves, a lot of events like the 2003,

2006 and 2015 European heatwave, Balkan

floods of 2014, US heat wave of 2011 and

North American heatwave of 2018 have

been attributed to them. Research is

still being conducted on the definitive

influence of anthropogenic climate

change on natural atmospheric phenomena

like these. It shows us how small and

delicate changes in atmospheric

phenomena lead to catastrophic events

affecting the environment and society at

large.

>> Then there are transient components

where something you know even if you

have a string that's constant, if

something bumps it, it's going to

resonate associated with that. So the

combination of those two forced as well

as stationary waves is what these things

correspond to and that's essentially the

idea that we want to. Okay. So the model

here was the interaction of these

planetary scale wave jet stream dynamics

with local features determines the scale

persistent and location of climate

extremes. So the climate extremes I

introduced you to so far were the

freeze, the floods or storms and the

heat waves. So all of those in the mid

latitudes will now essentially be

related to this phenomena. And the other

point I tried to get across was that we

are used to thinking about what's going

on here. But what's going on here is

related to many other places because of

the same mechanism. So if we can mess

with this mechanism, we have the ability

to influence the whole wave train and

that could be good or bad because we may

affect positive change here but it may

end up with a negative change somewhere

else. So you have to think about how one

does that.

Okay. So this is just the illustration

of you know the same process we looked

at. So definitely move forward.

So this is uh now going to be a section

on atmospheric rivers which I got

introduced to in 1992 by a paper from

Shu and Noel Dual from MIT. They did an

analysis of satellite data and they

identified these filaments of water

vapor that were moving through from

basically five locations in the tropical

oceans. Uh and then it's like a hose. So

it starts here but it can spray there or

it can spray here because of the way it

intersects with the atmospheric

circulation. The instantaneous discharge

in these features like the one that

you're seeing spin out there is two to

three times the the typical discharge of

the Amazon River which is the largest

tourist field. So that's the importance

of these and there's an example of a

flood in California that happened

because of this. So the qu the the the

reason I put this up is that we are in

our group we are taking the atmospheric

rivers as the first demonstration target

for can we actually control what goes

on. So can we steer these rivers in a

direction by nudging somewhere uh and or

can we smear them out so that the

landfall is over a larger area so that

we reduce the flood potential but we

provide the water. So that's kind of the

goal

and here's another video. Let's see if

this will play.

>> In 1862, California was hit by a mega

flood as a result of continuous

atmospheric rivers. It rained for about

45 days straight in Northern California.

About a quarter of the state's real

estate was destroyed, and a 6,000 square

mile inland sea formed in the central

valley. The state's capital had to be

moved to San Francisco due to flooding.

Remember that California's population

was about 1% then what it is now. The

USGS in 2010 created the scenario for a

similar storm hitting California, which

happens on average about every 200

years. The report also found that a

similar storm to 1862 would force a

million and a half Californians to

evacuate and would cause more than a

trillion dollars in economic damage in

the long run. The report also outlined

which major areas of California cities

would likely be inundated and have to be

evacuated. findings included that the

majority of the people in the Sanwaqin

River Valley would have to be evacuated

in this scenario and it's possible that

an inland sea would form again in the

central valley as it has time and time

again. Basically what the what the story

is in that video is that around 2008 or

10 the US Geological Survey came up with

this idea of called the ark storm

scenario

u like Noah's arc and the story line on

this is the following uh there's a group

at the University of California Santa

Barbara sedimentologists

they started investigating sediments and

looking at uh the abundance of flooding

that happened at different times. And

this was motivated by the fact that

there was an event that this video is

talking about in central California

where much of central California,

central valley was flooded for about um

the whole season basically uh

continuously in 1862 and Sacramento had

to be evacuated. the state of California

government moved to San Francisco and

their argument is that if the same event

were to happen today, you would have to

evacuate one and a half million people

from that area and the losses would be

in the tens of billions of dollars. Uh

and it's not here that you know we could

restore it to conditions anymore. Um

so the USGS then looked at this and they

from the sedimentary analysis they

figured out that in central California

this kind of event has happened

on average once in 250 years. So it's

not again climate change but it is

something that we face. And um the 1862

event was marked by the fact that you

had streams of these atmospheric rivers

coming one after the other. So the wave

train was locked in position similar to

what happened with the freeze for seven

days but in this case it was locked in

for the whole season. So every four or

five days there was another atmospheric

river landing and so the supply of

moisture was continuous is the idea. And

so so that's kind of the motivation for

looking at this guy. Moving on to the

east coast. There our issue is uh

hurricanes

and there's been quite a lot of work

done in the past on how do I stop a

hurricane. So that work falls into two

categories. One is the hurricane is here

and the hurricane is sustained only if

the surface temperature of the ocean is

warmer than about 26.5° C. If it drops

below that, then there isn't enough

energy to sustain the hurricane. So

there's a bunch of people who've tried

to come up with creative ideas on how to

bring up cold water up or to otherwise

change the the thermal characteristics

of the system. There's the second major

attempt to diffuse a hurricane has been

to look at the wall of the hurricane and

to seed uh do cloud seeding kind of

activities there with the idea that by

reducing the temperature to changing the

temperature profile there you can blow

out the wall of the hurricane and you do

that. uh there has been modest success

with both of these efforts but the cost

is prohibitive and there was one event

in which things actually went a little

bit crazy so they stopped doing that

now the reason I'm showing you this is

my argument would be this these people

looked at the problem all wrong because

their goal was to diffuse the hurricane

they are thinking in terms of a science

experiment rather than protecting people

if the hurricane stays in the ocean. I

actually have no issue with that from an

engineering point of view or from a

social point of view. So if you look at

and I can show you hundreds of these

pictures with the hurricanes in this

particular case, this is this hurricane

and those are the jetream winds. So this

is going to actually be steered away by

those jetream winds in that particular

direction. So this is a outcome which

doesn't concern me. Here's another one.

different hurricane, same story. So in

the early 2000s, we built a stoastic

simulator for hurricanes uh using a mark

of renewal process and uh

it was intended to be a simulator for

you know just continuous simulation so

that you get a risk profile associated

with hurricanes. The student who worked

on it started running it on live

hurricanes as a forecast simulation

model. And my reaction was, "Are you

crazy?" And I was made to feel really

stupid when she actually predicted that

hurricane Sandy was going to have a

landfall in New York, which only one of

the 10 physics based models was

predicting. I was going, we have no

physics in this model. This is purely a

stoastic model. But as we looked at the

hurricanes and we looked at what our

model was doing, there was a selection

process in the model where if the

hurricanes that it was sampling from had

tracks associated with this and we went

and looked at what was going on in the

atmosphere during those hurricanes as

opposed to other hurricanes. It was

exactly the steering winds of the

jetream that were involved. So if the

jetream had been going way off there,

this thing has a landfall because

there's nothing steering it away. If the

jet stream is coming and curving way up,

you have a landfall going in that

direction. If the jet stream is doing

what it's doing here, it's gone.

So

now that's tying together all the things

I raised in the beginning as important

climate extremes in terms of losses and

all of them then have a connection to

how the jetream behavior works.

Okay. So I have thought through what we

are doing on climate adaptation because

this became a big thing because of

climate change and my answer is deadly

squat. Uh so we don't have a story. So

even if we you know I said if we

decarbonize we will still have these

issues and we actually have no plans a

lot of money is being spent on green

infrastructure natural infrastructure

all that is needed but most of that

doesn't address the things that I've

brought up uh it it doesn't address

losses in particular at the scale that

we are concerned about

um I talked about these and then in

terms of even early action uh the whole

issue of lack of pred predictability of

sensitivity to initial conditions gives

us very limited ability to predict that.

My colleague uh Dan Shra at Harvard was

trying to get me to show up today in

India because he's very concerned about

the heat waves in India and his proposal

was that I join him in talking to Prime

Minister Modi to make 1 billion people

in India live underground because he

doesn't see any other solution for them.

So think you know when if you I bring

that up because if you question the idea

of perturbing the weather you have to

look at the kind of things people are

proposing as an alternative to that.

Okay. So this is my argument. We need a

new 21st century approach and we are

arguing that this is whether jujitsu and

I will try to explain how that works.

So here's an idealized model which is

typical in any elementary chaos theory

textbook. What is it? What it is is the

three variables that you're looking at.

X is the mean velocity of the jet stream

and Y and Z are s cosine phase of eddies

or transient wave that intersect with

it. The way this border works is that if

you just look at this term here, dxdt is

minus x. So it's purely dissipative.

Okay. Uh depending on the magnitude of x

dxdt is negative and so it's going to

try x to zero. It gains energy from the

edies at some rate. Okay. What does the

edi do? The eddi is dissipative as well

and it gains energy from the jet speed.

So that's that gives you the nonlinear

oscillation that's become the canonical

model associated with this. So that's

the structure here. And you know this is

the typical butterfly that perhaps

everybody has seen. And as this thing is

going around what I'll mention is you

have these two areas which look like

almost circles. So there's an

oscillation happening in each one. It's

actually fairly stable and where the the

unpredictability or the divergence comes

is as you jump from one ring of the

butterfly to the other. Because think

about it, if you're traveling here, you

suddenly have the possibility of staying

with this travel or jumping to the other

side. That's as far apart as you can

get. And that's that's where the problem

comes with that particular model.

Okay. So the idea of the open exponents

can be explained as here. You have two

initial conditions there at delta 0.

They're separated by that delta 0 value.

Then as the system evolves uh these

trajectories diverge. So the rate of the

exponential rate of divergence of these

trajectories with time which is how

delta t grows as a function of the

initial delta o and the leoponov

exponent you see is delta t is delta o e

lambda t. So lambda is the rate of

growth of this divergence. Okay. So

that's how this behavior happens

overall. But we can actually look at how

that is changing locally over the

overall attractor. And so what you can

see is that there are places where

things are actually not at all

diverging. These are negative values of

that theoponomic exponent. And what you

can see is that those are at the edge of

their attractor. You can't just go to

space. You have to be constrained. So

you come back and then this is the area

as I said earlier where you have the

highest values of the geoponics.

So if you think about this in terms of

sensitivity to initial conditions, these

are extreme values here, I actually have

pretty good predictability there. Where

I don't have predictability is when x is

approaching zero. So I have sort of an

unstable situation. Okay, very

interesting to think about in that way.

Okay, so there are local eopanov

exponents which are calculated at a

particular position in phase space and

there are finite timeov exponents where

I can look at where I am today is

closely related to the eoponov exponent

but I look at what my neartime evolution

might look like if you know the

equations this you know fairly

straightforward to do the second model

that Lorenz introduced that's not as

common in the chaos literature but is

you know foundational in the meteorology

literature is the Lord's 1984 model.

It's very similar to the other one. The

main changes are these FNG terms that

have been introduced. This represents

the equator to cold temperature gradient

as a forcing to that jet stream and this

represents the land ocean temperature

contrast as a asymmetric forcing to the

edies basically but otherwise the model

is fairly similar. So if we want to

think in terms of this this idealization

of the jet stream because essentially

the rosby wave dynamics are what these

models are generating. This gives us a

way to think to then think about how we

could introduce our pertations and the

perturbations can be introduced in the f

terms which are the boundary forcing. If

I want to study the behavior of El Nino

together with a jet stream, we have

published on that and we can take

idealized models of the El Nino southern

oscillation as a ocean atmosphere

interaction and then use that to force

FNG in this particular model and you

have a couple model idealized coupled

model that go through. So the idea now

is to do adaptive chaos control. And so

what the gojitsu part of the argument is

that this thing is is losing

predictability or deforming anyway.

Let's just use that. We don't have to do

a whole lot of work to do it. So if you

think about the hurricane example that I

gave you, all the people who worked on

the hurricane worked on the hurricane

and at that location.

The argument implied by this is that if

I want to change the jetream, I'm

probably going all the way to the

Pacific for an Atlantic hurricane to

induce the change way upstream, which

then propagates forward over time and

changes the dynamics and it would be a

small perturbation of some sort. So the

adaptive aspect of it is that as I

introduce these perturbations, I also

have to see what happened and then I do

an update through data estimation and

move things forward. So that's basically

the game

and the question is can we control the

trajectories for the Lorenz 63 or 84

model and Moan who's sitting right there

and Shin who is not the ones oh there

she is yeah you're hiding behind the

other person so so they've been working

on this and so basically what is what

they have done is they have set up a

optimization model just to see the

feasibility of the game so far so you

apply a perturbation but you want to

minimize is the total energy that you

are putting in and subject to

controlling the trajectory of just the

jet stream. In this particular case, you

could expand that and uh so you you look

at a time forward and you say I want to

keep it right there and you know see if

that's possible or not. Um and yeah and

the energy is defined in terms of

perturbations in X Y and Z. Those are

the decision variables.

So here's their result for the Loren

model and what they specified here was

that they restrict the trajectories to

stay in the positive quadrant. So it's a

pretty broad zoom. It's not, you know,

restricting to a fine window. Uh left

side is the natural evolution of the

the load 63 model over 2,000 time steps.

The right hand side is the control

version. So you can actually do that and

um you can look at also cases where this

fails.

So it depends on the initial condition.

You aren't always going to be able to do

this. And learning when you can and when

you cannot is part of what one needs to

explore. And then here what he's got is

the total energy of

the associated exponents and they match

which is what sort of what you would

expect. And then what's the percentage

of the energy of the system that you are

actually working with? And you can see

that those are 0.02%. This is very

small. So that that was kind of the

hypothesis associated with trying to

play this game. And the lens 84 model

the same game and a similar result

associated with that. And then in this

case the perturbation that you end up

having to apply is quite a bit more.

It's starting to approach 1% associated

with it. But uh there's if you if you

look back at the journey

associated with the attractor that has

been suppressed it's also quite

substantial.

So this is how far they have gotten so

far and what we are shooting towards is

a more first of all a more realistic

modeling strategy. So we take high

resolution historical weather forecast

we forecast data on a bunch of bunch of

attributes. We don't want to run a GCM

because there's simply no way to run do

an optimization on that. So we use an a

IML space-time emulator that's a

pre-trained model. So you basically

you're just calling it there and then uh

you do an ensemble with that. Uh you run

the optimization you look at the

feedback and basically see we can

actually control these things and the

target is the atmospheric rivers in the

Pacific. And so that that has that's a

data set that's been processed and it's

available to us at this time. And what

you're solving for is different from

what we did with the lower in this

location

and timing of the perturbation. So in

the previous one we allowed a

perturbation at every time step. Now we

are wanting to minimize the number of

interventions that you also do in

addition to everything else. Um and uh

then to steer this or to precondition

it, we would calculate these finite time

or exponents and use them to provide a

prior distribution for when to do these

perturbations and how much they should

be. And there will be a regularization

aspect to this as well. So that um

you're you're you're not doing this kind

of behavior all the time. So to

kind of lend credence to the argument

because we haven't done this yet. On the

left is a big snapshot of atmospheric

river in the Pacific. That is from this

paper down below. And what these people

did is they calculated the finite time

yonog exponents associated with the

whole picture not just the atmospheric

river. And you can see the atmospheric

river actually pops out. It's a local

maxima in all the geopolic. So what that

says is that a small perturivation will

grow quite rapidly in that particular

area and hence this is a good place to

start messing around with rather than

trying to move the mountain.

Okay. How do we you know every time I

give this kind of a talk people say how

are you actually going to provide the

energy to change this? One of the

colleagues here said oh you're going to

blow up hydrogen bombs to do this. I was

going that's not the idea. Uh just to

clarify where he got that idea is that

the people who were trying to control

hurricanes, they calculated the total

energy in the hurricane system and then

they calculated the total energy it

would take to to squash that and it was

in the tens of terowatts to do

something. So that was hopeless here. We

are not trying to do anything like that.

So we have to still think about what we

do. And so one of the thoughts is that

if I look at a typical thunderstorm,

what's the amount of energy dissipated

by that thunderstorm? It's typically in

the order of 10 gawatt hours. So it's

substantial. Basically, uh if I do cloud

seeding or other interventions to

generate a rainfall, I need a lot of

humidity.

When people try to do cloud seeding in

the desert to produce rainfall, there

isn't humidity. They're trying to force

the issue. This is a different

situation. I'm looking at something

that's carrying the most moisture

possible in the atmosphere. So

intervening in that particular way here

is possible and you're doing that

intervention over the ocean is the idea.

So so there are several things some of

them just sounded completely wacky to me

but I've included them. There's a group

in China that has been taking uh large

speakers, audio speakers, and pointing

sound waves at the atmosphere. And they

claim that there's a 17% increase in

precipitation by putting u 50 decibel

noise up. Uh I would love to see this

happen.

We were criticized by one of their

colleagues who said that these people

need to replicate their experiment multi

hundreds of times before they will

believe it. So they have published three

papers since then showing statistical

evaluation of their results and they

claim to do it and uh they they provide

some physics based arguments. There's

people at the University of Geneva

who've been doing these laser

experiments

uh and these are high-owered lasers. So

what I found really fun about their

story was they point the highowered

laser which attracts all the lightning.

So they focus a lot of energy into a

small area and then they use low powered

lasers to detect the the collisions of

the water molecules to figure out what

the induced rain rate is. So so you know

I'm gone that this is pretty exciting

stuff but I don't know if any of these

things will work. The third idea that's

been pointed out is space-based

microwaves. So you can actually generate

enormous amounts of energy and actually

focus it very narrowly if you wanted to

do that. Uh and the nice thing about

that one uh if it were practical is that

if I have stationary satellites in

geostationary orbit, I could basically

do this wherever. So if I need to do an

intervention here and the next

intervention is half the way across the

Pacific, it's doable. So that's kind of

the idea. So to summarize, what's the

potential? I I don't know. This is an

idea that I've had for some time. It

opens up a new area of research for

people which is a different way of

thinking about it. I do think there's a

need for developing weather control

technologies. Theoretically it seems

possible. We have a specific first

target. I would love to do hurricanes

but I want to see if we can do something

with the atmospheric rivers first. So

what we our goal is to first demonstrate

this with numerical modeling of the

realistic case and then in parallel work

on understanding what are the physical

mechanisms for perturbation. Who else is

doing this? Unfortunately not very many

people. There is a project in Japan

called Moonshot 8. So Buaan and Shin

located these people at EGU zoomed in on

them like vultures and now we are

working on a collaboration with them.

They got I think $2 billion for this

wonderful project. I have a feeling

President Trump will be very happy to

give us the same amount of money

and UAE has a little bit of work and

China has some work and that's about it

at the moment.

So, thank you for listening. Uh this is

the story line and I would love to see

if anybody wants to get involved.