Climate Change in 4 dimensions

8/12/2019 Climate Change in 4 dimensions

http://slidepdf.com/reader/full/climate-change-in-4-dimensions 1/36

Okay today we're going to talk about computer models of the climate system and I like this

topic because I'm a climate modeler myself so what I do for my own research is work with

these models and try to make them more realistic and try to use them to learn more about

the climate system. So when you think about computer simulations I think it's important to

realize that we're living in a time when computer models are everywhere. The airplane youfly in was largely designed on a computer, the car you drive used computer simulations

throughout the design and manufacturing process. And you may notice there is not very

many tests of nuclear weapons, certainly not in the major countries anymore. All the tests

that used to happen on land and in the ocean and so on in the air were banned and one

reason why the ban was acceptable to nations was that they had already learned that the

computer simulations were extremely informative and in many cases my friends of mine who

work with nuclear weapons say we got to the point where when we set off a device to make

a test it just reassured us that the computer simulations were accurate. And in science,

computer models, first of all, they're new. Big computers have only been around for about

sixty years and secondly, they're especially valuable in fields like climate, where it'simpossible to do controlled laboratory experiments. You can't take the atmosphere and put

in a test tube, heat it up, cool it off, change its composition. The same is true for the ocean

and lots of other areas of science when you think about it. Geologists can't take an

earthquake and put it in the laboratory. And astronomers or astrophysicists can't take a star

or planet and do experiments on it. So it's in fields like that, where a lot of the researchers

observational, rather than laboratory oriented, where computer models have often been

really revolutionizing tools.



So I'm going to start with a bit of history. We mentioned all of these people the last time we met. I didn't say

very much about Fourier, the person on the left, but Fourier was a brilliant mathematician and physicist. We

owe enormous progress in both physics and mathematics to Fourier. He was especially noted for his treatment

of heat transfer. And you might say although it was in many cases abstract and to a degree speculative or

conjectural he laid the foundations for what we now understand when we talk about things like the energybalance of the earth which we introduced last time We said more about the other people Tyndall, the second

person, it says English, it's true he worked in London. All his scientific research was done at the Royal

Institution of London, but he was Irish, he was very much Irish, he grew up there and Tyndall was the person,

you remember, did the laboratory experiments that demonstrated or found or established that carbon dioxide

and several other gasses are heat trapping, they absorb infrared radiation, they're responsible for the

greenhouse effect. Arrhenius was a chemist, and brilliant chemist, who later won the Nobel Prize and he, in a

sense, did the first calculation of how much you might expect climate to change if you change the amount of

carbon dioxide in the atmosphere, either raised it or lowered it. And as we mentioned, his estimate for us that

if you doubled carbon dioxide you'd warm the world by about 5 or 6 degrees Celsius, 9 or 11 degrees

Fahrenheit, roughly. That estimate is within a factor of two of the modern ones, which we're going to discussin a few minutes. Callendar was, in some sense, a predecessor of Keeling. Callendar was a British steam

engineer and unlike Keeling, he did not invent an instrument and make measurements of carbon dioxide, but

he made use of what observational data there was and he conjectured, he speculated that carbon dioxide was

arising in the atmosphere and he was a very serious scientist who really did make the strong case that if the

carbon dioxide were increasing amounts in the atmosphere as fast as some of his observational data

suggested then it could impact climate. And Keeling, as i said, put this field, observed changes in carbon

dioxide, observing the amounts in the atmosphere on a firm empirical foundation, meaning many ways what

he did was akin to what Tyndall did. Tyndall built his instrument, made the measurements in his laboratory,

showed that carbon dioxide absorbed infrared energy. Keeling designed and built the instrument that

measured carbon dioxide in the atmosphere, installed instruments at places like Antarctica, and especially theMaunaloa record in Hawaii which gives us this very clear, unassailable record, the Keeling curve of rising

amounts of carbon dioxide, from 1914 or 1958 until the present.



s today, I want to introduce another historical figure that we talked about in the readings for today. okay we have our first

omputer glitch or reading or impressing advancement doesn't do that reasonably related events is funny 'cause this time we

pend a lot of time before class making sure it was exactly right This is the person I wanted to talk about today, his name was

wis Fry Richardson he was born in the 1880s, he died in the 1950s and he was a Cambridge educated mathematician andhysicist. He made important advances in other areas of science and he was an early enthusiast for numerical methods for

bsolving differential equations. There's many equations that you can write down, but you can't solve analytically. They come

om various parts of calculus for example and it's possible to essentially convert them to arithmetic problems so that if you

n do a large amount of arithmetic you can make an approximate solution to the equation and in a sense, he was interested

weather forecasting, not climate at all but weather forecasting as a possible application for this kind of technique. Now min

ou there was no digital electronic computer in Richardson's time, he first got interested in this a hundred years ago. This year

013 is really the 100th anniversary of Richardson beginning to think seriously about how you might make weather forecasts

y a method that was more scientific than what had previously been the practice. In Richardson's time a hundred years ago,

eather forecasts were made by the rules of thumb, by little semi-quantitative reasoning There wasn't very much

bservational data so even if you knew that you should watch for the storm that would blow in from the west, you might not

ave very much information about the west but the data was beginning to become available. And Richardson's view was whyn't this be formulated and solved as a scientific problem exactly the way that we can predict tides or eclipses that's not done

y rules of thumb and today we don't think it's remarkable at all that you can look up on the tide table when the beach will be

e biggest and smallest and go to the beach and there it is, high tide and low tide come as expected. We don't think it's odd

at you can read there will be a solar eclipse at 10 o'clock tomorrow morning and you go outside and there it is. And

chardson felt the same way about the weather, he said we ought to be able to write down equations that express the

hysical laws, conservation of mass, conservation of energy, and so on, that govern the weather. After all there's nothing

ccult about the atmosphere it's just a gas, we know what the equations of motion for gas are, it's Newtonian physics, it's

assical and if we observe the initial state we have to be able to predict so if we know what today's weather is and we know

e physical rules that the atmosphere follows, we ought to be able to predict tomorrow's weather. In this case by whether he

dn't really mean where it was cloudy or not in La Jolla, he meant the large-scale motions of the atmosphere, the things you

e on the weather map on television where the highs and lows are, where the cyclones are and so his goal was to basicallyrmulate that as a scientific problem and solve it using his equations and he did this work in very strange circumstances.

chardson was a Quaker, he was a pacifist, and we're talking about 1913, on the eve of World War I and when World War I

oke out, the following year Richardson because of his convictions couldn't join the army or take part in combat but he was

ery patriotic and he was also very curious to see war close up and he was like many people a little bit frightened, he didn't

now how he'd react to it so he tried to find a way to get involved in the war without being in combat. And after several

tempts he got assigned to an ambulance unit which was in France and was bringing wounded soldiers back behind the lines

om the front where they got wounded and so he did this for quite a while and meanwhile he was thinking about using

omputing to predict the weather and in-between trips to the front he actually carried out calculations he built the first

eather forecasting model which I'll tell you about in a moment. He carried out calculations, again there was no computers, s

e was doing pencil and paper arithmetic and it took him weeks and weeks and weeks to make his first forecast and it was a

rribly wrong forecast, but he learned a lot from the experience and he lost his papers, they got found later, and he published

em in a book in the 1920s which was in a way the first textbook in modern meteorology and we today are using methods

ery much akin to what Richardson thought up. There's been advances, there have been progress in many ways, but he really

as a pioneer, he was a visionary and I'll show you his idea.



This is a map of Western Europe you can see the British Isles on the upper left, you can see the bootof Italy in the lower right and what you see there in this checkerboard pattern is this map with squaresof about 200 km on a side laid out and little labels in the middle. The labels in the middle were thenames of towns Richardson thought that it was a great shame that weather observations, be it withbarometers and anemometers and wind vanes and so on thermometers, that they've made thosemeasurements in prominent cities, capital cities in big cities and so on. He thought it would be muchmore logical if the weather stations were arranged in a regular pattern like this so that you hadmeasurements of today's weather on a regular grid and his ideal was the following, and we canexpress it mathematically but it's also quite possible to express it really with just words so for examplesuppose so you were interested in forecasting in forecasting what the weather was going to be at thissquare here and for concreteness, let's say this square is near Munich in southern Germany. Supposethe wind was blowing from west to east that means that this square here would change in it'stemperature really want to make this cursor show up when it comes This square would change itstemperature if the wind was blowing from the west to east and it were warmer in the west then thethe wind will blow warmer air into the area you're interested in and the temperature would rise. Andthe temperature would rise more if it were much warmer to the west and it would rise less if it werenot so much warmer, it would cool if it were colder to the west and the stronger the wind blowing thewarmer or colder air in, the bigger that effect would be. So what we're talking about in jargon isadvection. One of the reasons the temperature changes is that we can blow in warmer or cooler airand so he thought if you had observations like this at all the squares so you know what thetemperatures were you could compute the effect of the wind blowing warmer or colder air in. Andsimilarly for all the other processes that he could think of he tried to express them mathematically inphysical formulas that were mathematical expressions of physical laws. So for example, if the sun wasshining brightly in the square you were interested in, that would cause the temperature to go up andyou could compute an amount depending on how high the sun was and what season it was and so on.And if the sun was shining, but there were clouds there, there would be less so to compute the clouds,you'd have to compute the humidity and so on. And he thought up of the elaborate program, carriedout calculations for making a 6 hour forecast at one of these squares. Their forecast was terriblywrong, was hundreds of times bigger changes that could ever happen, but he knew why he thoughtthe initial data weren't good and in retrospect we know that that was true. After all there were nodata at all in the upper atmosphere he was just relying on surface observations and he also was using acomputational technique that was flawed because he was an early pioneer in that too. Butnonetheless he persevered and he thought of the following vision…



Imagine an amphitheater like this with thousands of people in these rings, each representing one of

the squares and these people would carry out these calculations because the idea was you predict the

weather a little bit ahead, then use that as the initial state to predict the weather a little bit further

ahead of that. Your neighbor wanted to know what your answer was because that defected what

happens when the wind blows there in from neighboring squares and so this guy at the top with aflashlight is shining a green light on people who need to speed up and a red light on the people who

need to slow down, we all have to proceed at the same time. And there's thousands of people, he

computed how many people he thought it would take to keep up with the weather because he

realized you couldn't take months and months to crank out a forecast for one day if it was going to

useful and he came up with a number of several tens of thousands of people, but he said if the system

could be made to work if the data were good enough, if computational techniques were good enough,

if there's people organized well enough, then maybe the benefit to the economy of the world would

be greater then the cost of taking the observations and hiring these essentially computational slaves

or graduate students. And so that was his dream and in a way what he's got here in his vision which he

expresses very eloquently in the book is the first design for a massively parallel computing system and

all these people are doing similar operations at the same time. Weather forecasting by these

numerical techniques became practical, it was really carrying out Richardson's dream.



This was the computer that was first used to make weather forecasts. Again, you're not predicting rain

and snow, you're predicting large-scale motion of winds, pressure fields, temperature fields, and so

on. And this is the ENIAC computer in Princeton that was used in the late 1940s. By the late 1940s,

everything had changed. The first electronic computers have been invented, this was one of them,

observations have become much more plentiful, largely due to the demands of military aviation. Theobservations that were carried out, especially during World War II because unlike in Richardson's day,

military aviation was important so it was important to be able to observe the state of the atmosphere

high above the surface and there were a lot of advances made in writing more realistic and tractable

equations than Richardson had and devising numerical techniques to solve them. And so the first

forecasts of the evolution of the large-scale wind and pressure and temperature field for a period of a

few hours were made in the late 40s and they were successful. They were as good as the best

subjective forecasts made by experienced meteorologists who were just looking at whether maps and

using simple calculations and rules of thumb and back of the envelope kind of thing. And by the 1950s

in fact, all of the rich countries weather service's had set up weather forecasting systems based on this

work which was, in a sense, a vindication of Richardson's dream. Today this computer here, which had

thousands of vacuum tubes, is far less power for them the computer in your cell phone and in fact one

scientist whom I know recently programmed a cell phone to carry out this calculation to repeat these

1940s achievements and it took many hours on a 1940s machine and it takes about a second on a cell

phone but using the same program we get the same answer One of the things that have happened

between that time and now is that computer speed has increased as well as our understanding and

ability to write realistic equations and to observe the atmosphere well enough to have the initial state

for forecasts.



Figure 1.2. The complexity of climate models has increased over the last few decades. The additional physics incorporated in

he models are shown pictorially by the different features of the modeled world.

Okay that's the end of the history lesson. This slide here I showed last time, but I wanted to remind you of this very rapid

evolution of climate models. The first climate models, as opposed to the first weather forecasting model that I've just been

alking about, the first climate models were put together in the 1960s and 1970s. And the idea then was not so much to

hink about how CO2 would affect the climate if you and I put it into the atmosphere. The idea was really simply a research

ool, in fact the first of these models were called general circulation models and general circulation is a jargon term in

meteorology which means the large-scale time average wind fields. So for example the trade winds that blow from east to

west in the tropics are part of the general circulation. So the question was interesting theoretically was how can we explain

hat? And the answer was well maybe we could take a model like Richardson's and instead of running it for a day or so of

imulated time to produce tomorrow's weather forecast, maybe we could run it for months or seasons or years and look at

he average of the atmospheric circulation and maybe we could do experiments again, thought experiments that we can

do in the computer that we can't do in the real world maybe we could take the simulation in our computer system and

make the earth rotate faster or make the earth have all mountains and no ocean or all ocean and no land or change the

trength of the sun and see how these factors that might influence the way the circulation of the atmosphere behaves,

how they react to changing external stimulus's or stimuli and the parameters you might say there is distance from the sun,

he composition of the atmosphere, and so on.

Lesson 2

So originally, these models, general circulation models, or GCMs, were just abstract research tools allowing people to do

hought experiments in a way similar to the kind of thought experiments that scientists have always done. Einstein had

hought experiments in which he imagined he was traveling along at the speed of light and so on and asked how the

phenomenon would behave. So in the same way you can do a thought experiment on the large-scale time averageirculation of the atmosphere, which you can't put in a test, but if you have a computer simulation you have a chance to

ee how it reacts to changing things So historically, that's how the models that we're talking about now came into being

and beginning about the 1970s, the late 60s and the 70s, it occurred to some scientist, maybe we could build a model

model that would let us see what the effect of changing greenhouse gas allowance would be on the climate. We had

Arrhenius's result from way long ago, but much more was known today about that kind of thing and suddenly we have this

potential computational tool. So these models, still called GCMs, now standing for global climate models, instead of

general circulation models. These models began to be developed and began to be made more and more physically

omprehensive and sophisticated. And so here's again this time sequence that we looked at briefly last time. At the upper

eft, the first models were just models with an atmosphere very much like the weather forecasting model running for a

onger time in which you could arbitrarily change the amount of carbon dioxide in the model. There was a simple

hydrological cycle Rain was falling depending on how moisture was affected around and as time went on more and morephysical complexities were added here and I'm not going to read this slide to you again, we saw it last time. Until, if I look

at the lower right the model in the AR4, remember the fourth assessment report that 2007 reported the

ntergovernmental panel on climate change, now had ocean models coupled to the atmospheric models..



Figure 1.4. Geographic resolution characteristic of the generations of climate models used in the IPCC Assessment Reports:

FAR (IPCC, 1990), SAR (IPCC, 1996), TAR (IPCC, 2001a), and AR4 (2007). The figures above show how successive generations

of these global models increasingly resolved northern Europe. These illustrations are representative of the most detailed

horizontal resolution used for short-term climate simulations. The century-long simulations cited in IPCC Assessment Reports

after the FAR were typically run with the previous generation’s resolution. Vertical resolution in both atmosphere and ocean

models is not shown, but it has increased comparably with the horizontal resolution, beginning typically with a single -layer

slab ocean and ten atmospheric layers in the FAR and progressing to about thirty levels in both atmosphere and ocean.

There were also models of ice at the land surface, of biogeochemistry of detailed cloud processes More and more, you

might say physical complexity and the models that potentially have the ability to simulate more and more of the processes

that are important to the actual climate system. But this is the evolution that went on, you might say in parallel with the

climate actually changing. I think, in a way, it's remarkable coincidence that the same time that the climate was changing fo

the first time due to human activities, we achieved the ability to observe it better, think of satellites, and also to simulate it

in a way that hadn't been possible. Before I could have easily imagined a world in which climate was changing, but we

hadn't developed these advanced technologies to understand it.

I showed you this picture last time too. Remember the jargon here, FAR, SAR, TAR, AR4, stand for the first and second

assessment reports of the IPCC. They came out in the 1990s. TAR, the third, and AR4, the fourth one, came out the firstdecade of this century. The fifth assessment report is due out in September 2013 and later. And there you see the grid

equivalent exactly to Richardson checkerboard pattern getting finer and finer. Richardson picked 200 km basically on

intuition and at the time remember he was working in the first world war in 1914 time period and when the first

assessment report was written, the climate models were using grids considerably coarser than that. So on this top picture

you can see the Mediterranean sea being about six grid points, whereas simply the increase in computer power by 20 later

at the 4th assessment report, you saw the grid is now down to 100 kilometers lower, that is finer by about a factor of two

than the 200 kilometers of Richardson's checkerboard pattern The weather forecaster is still made using these kinds of

models. You don't need a model of the the ice and the land surface of the ocean if all you're trying to predict is tomorrow's

weather but the atmospheric model and today's operational weather forecasting models in this country, in Europe, and

elsewhere have grids off in the neighborhood of 10 km that can be very fine. It's coarser for a climate model because you're

solving more equations as more of the climate system is involved and you're running your model for simulated seasons in

years, in decades, and centuries rather than from just today, tomorrow, or next week.



So if you look and see what needs to be done to make climate models even better, the causes of theunrealistic features, or if you like, the causes of our uncertainties, that causes that lead the modelresults to be perhaps less realistic, only a few, at least the ones we know about. There's always apossibility of things we haven't even thought of, but we know already that clouds and their interactionwith both sunlight and radiant energy from the earth, infrared radiation, are a huge effect. And lasttime when we talked about the energy balance of the earth, remember clouds are responsible for abig fraction of the earth's reflectivity. Roughly 30% of the sunlight hitting the earth is reflected away,bounces out to space, doesn't affect the climate at all. And of that 30%, 20% is being reflected fromclouds, the other 10% from ice and snow and bright things on the surface. So clouds dominate theprocess by which the earth reflects away a lot of sunlight, therefore cools the earth and as it says herethey also contribute to the greenhouse effect. And as I may have mentioned, these are things that areexplainable to little children in everyday terms. Little children in La Jolla know if you go out on thebeach on a cloudy day, you're cooler than on an otherwise similar sunny day because the clouds havereflected away some of the sunlight that would have come through if the sky was clear. And if you gocamping on an overcast night, you're warmer than you would be in a similar clear night because theclouds are like a blanket there, adding to the natural greenhouse effect. And globally these areimportant and in fact, it was a very interesting question to ask and scientists have asked it for decades,what's the net effect of clouds? That is to say is the cooling stronger than the warning or the other wayaround? So if you imagine, another thought experiment, a planet just like this one, but with no clouds,would it be warmer or cooler? And we didn't know and there were all kinds of theorizing about it. Weknow now, because of satellite measurements, satellites in the 1980s were sent up that could actuallymeasure the strength of the radiation from the sun that was bouncing up from the clouds, thereflection or albedo part, and the infrared energy that the clouds trapped and the answer was thecooling was stronger. They were both very big, but the clouds had a net effect globally, averaged overthe years, of cooling so that if they were all to go away you'd have a warmer planet. That doesn't say,however, how clouds will change if the climate changes and the short answer is we don't know. It's ahuge unknown. We know more than we used to, but we still can't say globally that in a changingclimate will there be fewer clouds, more clouds, higher or lower, more full of water, less full of water,therefore how they would affect their radiative properties, their reflectivity, their infrared absorptionof clouds. And different also made different treatments of these clouds all by smart well-meaningpeople, but getting different answers and that leads to a spread in the result.



And I want to introduce one jargon term here, which is climate sensitivity and it's the thing that Arrhenius first tried to

get a handle on. So we did the thought experiment, you have a powerful god from mars, you come along, you dump

twice as much carbon dioxide into the atmosphere is a you dump again as much that's in there so now there's double the

amount and then you go back to mars, fold your sixteen arms and wait and ask yourself the question what will the earth's

climate do with double the amount of CO2. Will it warm up because you strengthen the greenhouse effect, but will

feedback processes change that? So for example in the warmer world, will there be more water vapor evaporated from

the ocean? If so, the water vapor is a green house gas so that could add to the one end and be quantitative, how much

will this warm it? Does doubling CO2 warm it about one degree, or three, or ten, or hundred degrees and just how much

does the water vapor amplify it, if at all and how does this vary with season and all the rest of it. And i've mentioned in

the middle of the slide here that there are questions, this is not a simple definition, it's a thought experiment but if you

want an exact answer you've got to really give the conditions of the experiment. Once again this whole powerpoint showwill be on the class website you'll get to see it but, when I say what processes do we allow you can think of lots of

processes and when there's research being done on one, such as if it warms enough, will methane that's now trapped in

frozen form, deep in the ocean or in the land or in the arctic be released methane's a powerful greenhouse gas and

there's geological evidence and several other reasons for thinking that you can cause methane to go into the atmosphere

if you warm the arctic enough. And so that's a feedback process that might lead to an increased warning if you waited

long enough for it to happen. Or another one, how long will we wait, we know that, for example, in geological history

when the world was only a few degrees warmer than it is now, the ice caps on Greenland and Antarctica largely melted,

but we don't know how fast that happened. And prevalent thinking is that it probably took centuries. So do you wait

centuries enough for that to occur? What kind of timescale did you have in mind? And then there are a lot of other

feedbacks that we're less sure about. So for example, the ice and snow in the arctic melt, if they get warm enough, and

under where the ice and snow were, is darker ocean or darker land which is reflective and more absorptive so it adds tothe absorption of sunlight there and that increases the warming. It's as though you had your house wired funny so that

when it got warmer, it turned on the the furnace instead of the air conditioner and therefore amplified that warning. And

in fact, we now know, and we'll come to that in a moment, that that is happening in the arctic. We still don't know how

fast it happens and until very recently, we made serious underestimates of how fast that would happen. Nonetheless

with reasonable answers to all these questions and reasonable guesstimates for all these processes, the range of

conventional climate models today is that doubling CO2, waiting for the relatively fast feedback, like changes to clouds or

snow-cover, might warm the world, as it says at the end there by as little as 2 degrees Celsius 3.6 Fahrenheit or as much

as 4.5 degrees Celsius, two-and-a-half times that much. These estimates are constantly being refined, but we actually had

that range now for close to thirty years and we're not sure. The best guesstimate today might be at 3 degrees, but you

don't want to bet your last dollar on that. There's a certain amount of group think going on. If you build your brand new

climate model because you were very smart and ambitious and it came up with an answer of 2.8 or 3.3 degrees, you

might think, well that's great, I've got a terrific model here and it confirms the well-known result, but if it came out with

the result that was very different from those you might start looking for a bug in your code whereas in point of fact, that

bug was just as likely to be there in the code that gave the standard.



So this is an unknown to a certain extent, there's really a range here and we are upfront about the degree to which we're

not sure and that has consequences so let me tell you about a test that people who build climate models put them

through and that is replicating the way we know climate has changed in the past so what you're looking at here is the

global average surface temperature, that is, temperature the atmosphere at the surface of the earth averaged over the

whole world, averaged over seasons and these three lines here with the weird acronyms are three research centers that

keep track of this. There's no one instrument that measures the global temperature of the earth, instead you take all

kinds of data that are taken from weather stations, satellite data, ships of opportunity, aircraft data balloons and you do

quality control on it, you make assumptions about what data is trustworthy or not to make sensible assumptions abouthow you fill in data sparse areas and so these three curves don't look exactly the same, especially on the left you can see

way over here, early in the twentieth century they differed by a substantial amounts, this is in Fahrenheit degrees and so

whereas recently, last few decades of the last century, the first decade or so of the current century, the curves are

essentially on top of one another, and in fact, there's other groups that keep track of this too, and the quality control and

the data management issues have really been worked by very careful people, who know the error properties of the

instruments and so on. So we really have a pretty good record of the surface temperature of the earth and how it

evolved over the last hundred years or so. So suppose you take your climate model and started off with an estimate of

how things were in 1900, run it for a century, and you get a slightly different answer, or maybe a very different answer,

then you think, well, what I've put into this model as external input is what I think the factors were that changed during

the 20th century, so from Keeling's measurements and earlier extension of them from ice cores that I mentioned last

time, we think we know what the CO2 amount was, so you could put that in. But we know less about what the other

greenhouse gases were doing, the methane, the nitrous oxide, and so on, and collectively, they might have been

equivalent to CO2, especially early in this century. Also volcanoes go off, when a volcano happens, a large volcano, we

know now because they've been observed, can cool the entire global atmosphere by about a degree or to Celsius for

about a year or two before that material settles out. Essentially the volcanic eruption puts things into the high

atmosphere that reflect away sunlight, so it's a sudden pulse of air pollution that temporarily cools things and we don't

know exactly how many volcanoes went off a hundred years ago, or how much of what kind of material they put up. We

know more about recent volcanoes, but there's some slack there, some uncertainty about earlier volcanoes, and the

same for the sun. In the satellite era, we've measured how much the sun varies, but we only can guesstimate it from

earlier records from things like sunspots or tree rings in the early part of the records, so there's considerable uncertainty.

So I've just told you that different climate models have different sensitivities, that is, they react differently to these

external inputs and so you can tune these inputs, knowing the right answer, which is on this graph, knowing what the

actual temperature did over the 20th century you can, you might say adjust the forcings, that make your modelreproduce this, so that for example if your model were very sensitive to CO2, you might not have needed to have as big

a CO2 or other greenhouse gas increase and so on…



And that's not a very satisfactory state of affairs, I think of it as sort of the skeletons in the closet of

climate models, that this test that we put in a model to, can your model replicate the sensitivity of the

climate that we've now measured over the last century as represented by the surface temperature

record or is it harder to do and therefore do we have to treat the external parameters as forcings

because knowing the right answer any reasonable model with adjusted external parameters can dothat, but that's the state of affairs right now because we don't have a single perfect model that we

think exactly represents realistically all the processes and exactly has the same climate sensitivity that

the real atmosphere does.



I think you could say more things about climate sensitivity, I don't want to belittle

this, but in a way expressing it as one global number smooths over a lot of things and

so maybe you should think separately about how sensitive it is to changes in the sun

and how sensitive it is to greenhouse gases…



And similarly, aerosol, which is jargon for small particles, pollution, aerosol in

everyday English means spray can, but in meteorology, aerosol means small particles,

liquid or solid particles, so dust in the air is an aerosol, pollution from a smokestack is

an aerosol, and so on and these things are highly local, unlike the CO2, which stays in

the atmosphere for centuries, and is mixed around by the winds, the amounts are

pretty much the same everywhere for climate purposes, aerosol has very local

effects, so its most polluted near the source of the pollution and the aerosol might

settle out or be rained out in a week or two. But perhaps cloud effects are like that

too so perhaps even if globally the clouds, which is as we said both cool and heat,

maybe if the feedbacks tend to wash out over large periods of time, it still might be

locally important so there's a lot to be learned there and I want to point out that

when we talk about the imperfections in climate models, that's the sort of thing we

had in mind.



Okay, this figure was in one of the readings, the article by Rahmstorff that all from science in 2007.

Just want to reiterate what was said here, which was he said these are looks at how the earlier climate

projections by IPCC reports up through the third assessment report panned out when compared to

observations and the answer was that, in general, they were doing well carbon dioxide was going up as

had been anticipated, so was global average temperature, increasing by about two tenths of a degreeCelsius per decade, but other features like sea level rising at the top end of the range of previous

projections, so we concluded in that paper that earlier projections as summarized in the IPCC report,

remember IPCC doesn't do research but IPCC authors evaluate the published research and come to

judgments about what's known and the judgment there was that the IPCC summaries of projections of

sea level hadn't exaggerated sea level rise and may well have underestimated. We'll say more about

that and a couple of the projections in a moment.



This we looked at last time, I only put in here to illustrate another very powerful use of climate models.

Yes, there is a question. What is the data relative to, is that arbitrarily 1990? Yes, that's right, so we're

just looking at sea level. At left hand part, are you looking at the bottom one? Where sea level and

temperature are. Yeah that's right so that's where an arbitrary zero in a case of sea level and for CO2

concentration, we give you the actual numbers, but in the other two, temperature and sea level, we're just giving the departure from a reference state, which could be anywhere, so it's an arbitrary zero

there, which is essentially what it was in the middle of a graph which in 1990. Good question. So I just

put this in here to remind you again a very powerful use of a climate model is in what we call last time,

detection and attribution, detecting the climate change that's not likely to be due to natural variability

and then attributing it to a specific cause. If you remember this figure, this is the black line shows how

the actual measured temperature varied over essentially the 1900-2000 in the six inhabited continents

and, at the lower left, over the whole globe, lower-middle, over global land services that will overwrite

global ocean services, and the pink and the blue areas were the results of trying to simulate that to do

exactly this kind of simulation of the 20th century, pink and blue with different assumptions on the

model. Blue only including natural factors, volcanoes and the sun, and pink adding also the man-made

factors, including greenhouse gases and aerosols, so this is evidence that no climate model anybody's

been able to construct, replicates the observed changes without including these human caused or

anthropogenic man-made factors.



Okay here's another example of how you can use a climate model, again from AR4 the SPM-5 at the

bottom is, SPM is the summary for policymakers that you've read from the fourth assessment report,

and you're looking here at on the left side from 1900 to 2000, the actual 20th century observed rise in

temperature and on the right side, these are several possible scenarios here depending on what the

emissions were so they're labeled up here B1, A1B, and A2 are just jargon for low, medium, and highemissions and the yellow line, the year 2000 constant, is the hypothesis that if we could freeze the

world at year 2000 so from then on, we only emitted every year as much as had been emitted in the

year 2000, we would see that lower warming. So here you use a model to tell policy makers what is

likely to be the climate change resulting from several hypothetical scenarios of how fast we emit,

mankind emits greenhouse gases into the atmosphere.



We looked at these two briefly last time, here on the left in December, January, February, and the

right, June, July, August, though that middle scenario there is changes in precipitation. Brown means

less rainfall, blue means greater rainfall, and where there are dots, the models tend to agree more.

Where there are no dots, the models tend to disagree, so this is multimodel means that results of lots

of models are averaged together and where there are dots the models at least agree. Doesn't meanthey're all right, but, at least, in agreement where there's no dots, models gives substantially different

answers. so another use to climate models is predicting not just that time rate of change of a global

average thing, but predicting the geographical and seasonal distribution of the changes.



And here we also looked at these last time, I don't, again I don't want to belabor this, but again in the

middle column, changes in surface temperature in the decade of the 2020s is in the right-hand column

and the last decade of the current century, entering the decade of the 2090s, and you see the effects

of relatively weak emissions at the top, moderate emissions in the middle, and strong emissions at the

bottom. Right now, by the way we're following the bottom trail so far so emissions are rising veryrapidly, they are rising year over year with an exception of a single blip in the global recession, we're

emitting more and more greenhouse gases and extra emissions are largely coming from developing

countries with large populations, China, India, Brazil, Russia, which are industrializing on fossil fuels,

especially coal. Lots of jargon on this slide, at the top, AOGCM stands for Atmosphere Ocean Global

Climate Model, so it's a coupled model that's doing all these things.



want to show you some new figures that we haven't seen yet. These are taken from a report that's in progress that's out

n the web now, you can find it, it's out in draft form for public comment and it's called the National Climate Assessment. I t's

U.S. report so it's made by U.S. scientists with a focus on answers for the U.S. and sometimes even concessions to the units

mericans think in, like here degrees Fahrenheit, but and now instead of this emission scenarios, what the U.S. National

ssessment calls the representative concentration pathways, so I stress that this is a draft report, it might change then, until

etween now and when it's finished, but you see here much the same pattern here, the low pathway labeled RCP, which

ands for Representative Concentration Pathway, 2.6 is when assuming that there will be a rapid reductions in emissions,

hat is, assuming that the world goes green, because remember you can't change the U.S. climate very much just by

hanging U.S. emissions because the winds mix the stuff around so that complicates the geopolitics of it, but it means that

he climate change you experience in this country is the result of the average emissions put in by all the countries so there's

global commons issue to this here, but on the left there's a low pathway that means it's very substantial emissions

eductions, big emissions changes between now and mid-century. On the right, you might say business as usual, the higher

athway, RCP, representative concentration pathway 8.5, is continued increases in the emissions. So you see this huge

fference if you look at the key on the left, then you're seeing warming, and this by the way, what's being depicted here

as the temperature changes in the last thirty years of the current century, that is, 2071 through the end of the 2000s

ompared to the last three decades of the previous century, 1971 to 2000. So it's the temperature rise in a century and you

an see that on the right-hand path, there is, again this is in Fahrenheit degrees, they're warming is greatest over land, less

ver the ocean, therefore greater in the northern hemisphere than the southern hemisphere, greatest in the arctic because

f melting ice and snow producing a darker surface it absorbs more sunlight, this very strong vicious cycle, but there's a huge

fference there, it's just a huge difference of things that were 1 to 3 degrees on the left are 9 to 13 degrees on the right in

ound numbers, which is quite sobering and this is a result that's consistent with the results of earlier models in the last IPCCeport, there's many many models that have gone into this. There's a range, we saw it by the way, if you noticed on the left-

and column of this previous slide, so what you're looking at on the left-hand column is these are the the probability

stributions, let's see, here's for the central column that 2020 to 2029 and here in red, the probability distributions for the

ght-hand column, so what this says is that what you're looking at here is the peak of the distribution, that is the place

here most of the models agree, but the different lines here are the lines of the different models, so some models have big

ils out on the rights so this is you have to express this probabilistically, it's the most likely in terms of the average of many

odels, but there's a distribution of models and, as we just said, they have different sensitivities, different representations

f physical processes, and they don't all agree. But nonetheless, if you look at this figure here, the simple message it says is

hat the climate that our children and grandchildren will inherit late in the current century is within our control. That is if we

ontinue, we, mankind you and all seven plus billion of your friends, continue to increase the amount of heat trapping gases

hat you emit, you do change the climate of the future and as far as we know, you change it essentially irreversibly onuman timescales. We don't have any simple way to undo emissions once they're done. It's a very sobering thought and I

hink this, you might say this research doesn't change fundamental conclusions, but it sort of strengthens them as more

odels with deeper analysis and so on.



Again, these are model results. Here for example is the same thing for the U.S., what your actually

looking at here is the U.S. temperature, on average, is increased about a degree and a half since 1890s,

that's a degree and a half Fahrenheit, again this is all in Fahrenheit, and 80% of that increase has been

since about 1980, so it's increased rapidly in the last 30+ years or so and this is the further increase

that you would expect and again the big difference between lower emissions than higher emissions.Once again the regional details was whether Florida's really that much cooler than Ohio is less reliable

in the big picture, but with sufficient resolution, you can break it down.


http://slidepdf.com/reader/full/climate-change-in-4-dimensions 22/36 2

I want to say I worry about sea-ice, which is a very sensitive indicator and something very

sobering in terms of causing climate scientists to sit up and take notice, happened very

recently. The IPCC report, AR4, the last one came out in early 2007, working group on

physical science part, and sea-ice in the arctic varies a lot over the year. In the winter sea-ice

forms the area cover increases, and in the summer, part of it melts and the area decreases. And the minimum comes in September, and we've had in the satellite era, which means 1979

till today, good satellite-based estimates of how much of the arctic is covered by sea-ice, and

as you can see this yellow area here, plus the white area, is the average sea-ice extent at that

minimum in September of the year, averaged over the years from beginning in the satellite

years 1979 to 2006. In 2007, it was suddenly much less. So this is September 2007,

essentially eight months after the IPCC report came out, we suddenly saw a massive reduction

in arctic sea-ice extent.



I'll show it to you on a graph, this red line is the observed amount of sea-ice, basically including things

in the pre-satellite era, where they are less reliable. 2007 was way below, sorry, previous years and

way below the range of model estimates, which is this bluish area here with an average of the blueish

area. IPCC again doesn't do research, it assesses research that's been done and published, and so

people who study this produced models that made this estimate. Here, you see there is a reductiongoing on in time, this is up to 2050 and 2100 but we don't see anything in the simulated reductions in

sea-ce as the world warms anywhere near as drastic as what happened in 2007. And later on, it got

even worse, but that was a sobering moment right there that suddenly we realize that the models we

were using were probably too simple. The part of climate models that simulates sea-ice have to be

more complicated than just saying, well, below the freezing point of water at 32 degrees Fahrenheit,

it'll melt, and above, it won't, because it depends on things like the winds and the currents. Do they

move the ice sheet, this is floating ice, into areas where they are more or less susceptible to melting or

not, so it's a complex problem.



Again, this is a figure, in which sea level, now with tide gauges, the red old floats at the end of tears

like Scripps, pier methods supplanted by satellite observations that's really what we can measure the

distance from the satellite to the ocean. So you see in this curve in the upper right sea level rising

more steeply than before and at the high end of the IPCC projection range.



This is an interesting picture here, these are projections of the U.S. average temperatures. These are

temperatures in degrees Fahrenheit on the right or left, relative to the average from 1901 to 1960,

and that's for sixty years of the previous century. And the black line here is essentially what's been

observed and here you see that for a few decades in the future, doesn't much depend upon which

emissions scenario you follow. These are the older IPCC scenarios and these are the newerrepresentative concentration pathways, but they make a big difference by late in this century. It takes

time for this to happen so by 2100, you could with drastic emissions, that's this RCP 2.6, this dark

green curve here, have warming stabilized over the average of the U.S. now at four degrees, relative

to what it was in early 20th century, but otherwise if you let emissions continue to grow, it could be as

high as 12 degrees, so it's an average of the maps that we've seen earlier for the range of projections.

And although the models do differ in sensitivity, as we've seen, that is they react differently to the

greenhouse gas. The big difference here that you see on later in the current century between this and

this, that big difference is overwhelmingly due to differences in the emissions scenarios. So the models

differ somewhat from one another, but the way the temperatures differ between 4 degrees and 11 or

12 degrees warming average in Fahrenheit over the U.S. is basically do to whether you have rapid

reductions, serious reductions, for reducing the rate at which we emit greenhouse gases into the

atmosphere drastically compared to today versus business as usual and continued growth of the

emissions. The emissions are global, but you're looking at the effect on U.S. temperatures here.



Here is the same kind of thing looking at the range of sea levels so what you're looking at here is sea

level rising in feet. U.S. national claim assessment goes away from this metric stuff and uses feet and

Fahrenheit and so on that U.S. audiences are used to and again here's today early in this century this

little box here is blown up, here this is the range these are the observations, green is the satellite data,

blue is the tide gauge data, with rising, top of the range is higher now with these new emissionscenarios for high emissions rates. So that now, that by 2100, by 2100 the emissions the sea level rise

ranges from at the lower end here of 8 inches or so, two-thirds of a foot, up to more than 6 feet so it's

essentially of from a few tens of centimeters up to two meters or so and the most likely range is this

yellow bar here ranging from 1 to 4 feet. So there's a very considerable range in sea level rise and

there's an area where science is still you might say, in a state of flux, there's more than one kind of

model to produce estimates of future sea level rise. One technique is to try to estimate each of the

causes for sea level rise, melting ice on land, thermal expansion of the ocean, and another simply to

correlate it based on past data, how much warning of the past has raised sea level. One of the

advances that's happened since the IPCC report is that we now have measurements showing the ice

sheets on Greenland and Antarctica, which we hadn't known much about beforehand, are both

melting, caving icebergs, and adding to sea level rise.



Okay well here the same kind of thing for global temperature projections, you see the same kind of

range, depending on emissions if you look at the far right, low emissions the world might warm by 3

degrees Fahrenheit, high emissions by 8+ degrees Fahrenheit but 2100.



Okay and here again these kinds of maps, the low pathway with heavy emissions cuts on the left,

business usual on the right, and again you're going to see what you saw for the U.S. and what you saw

in the IPCC report, very different world in 2100.



And here again from the draft of the U.S. National Climate Assessment, these same kinds of things we

saw in precipitation changes, so if you look here, that the subtropics, which are already dried, the

latitudes that include deserts like the Sahara, (very sensitive computer today), getting drier, storm

tracks moving poleward, more rainfall in high latitudes.



I didn't want to spend a lot of time showing this, but I wanted to show you the resemblance between

the U.S. temperature changes that have been observed and the projections that we're making. The

point I want to leave and I'll be done soon so you can have time for questions and discussion is to show

you that there's reasons for taking these things seriously. I don't think it's always good to take every

computer projection literally, but they're quality of these projections comparing what we've observedwith what earlier models projected is now good enough that you can't ignore them, you have to take

him seriously. So let me decode this for you a little bit the map colors on this map of the U.S. show

that change in temperature in Fahrenheit degrees, and the color code is over on the right, between

1991-2011, essentially the last twenty years, last two decades, compared with the 1901-1960 average.

And you can see that in darker red there are changes greater than a degree and a half, again a degree

and a half doesn't sound like very much locally, it changes more than that between breakfast and

lunch here, but a degree and a half averaged over a whole area as big as these areas here, can mean

big changes in the likelihood of heat waves, chances of precipitation droughts, and so on. And on the

little graphs that you see surrounding the U.S. the bar charts, are essentially the changes by decade

relative to the 1901-1960, so you can see in every part of the U.S., the recent decades are warmer the

red bar is higher in all of one, two, three, four, five, six, seven, eight, nine, ten of these graphs then any

other bar. So in other words, each of the regions of the country here, the last decade basically has

been warmer than any previous decade on record. And the same pattern is what we see in these

projected emissions, so if you've toggle back-and-forth between these two maps, look at the right

hand map, the high is in the same graph we saw earlier there, that right-hand graph is the projection

for future changes you see it's less in the southeast and if you look at where the changes have been,

it's less in the southeast, it's great in the Great Plains and Great Lakes area and if you look at where it

is, it's also been large changes there, it's less in the pacific northwest here, less in the pacific northwest

there, so the details vary a lot, this map of observations is a lot patchier…

3



and this map of computer projections is a lot smoother, but you can't ignore it, the resemblance between

the patterns is strong enough.

3



Again here's the difference between the upper left, the rapid reductions in emissions, the lower right,

business as usual, and you see this huge change that would be expected by the end of the century. If

you're fascinated by these things, you can go and download the National Climate Assessment, it's on

public websites, just Google it and it'll turn up.

3



I didn't want to bury you in graphs, but this is an interesting one here. One of the projections of

climate models is that in a warmer world, we've seen more of the rainfall falling in heavy rain events,

less drizzle and more downpours, and what you're looking at here is separated by region here in the

U.S., is the percentage increase 1958 until the present, 2011 actually, of the amount of rainfall falling

and the heaviest 1% of all daily rainfall events. So if you look at the northeast there where it says 74%,what that means is in that part of the country, remember this is a very well instrumented high-tech

country, so we've got reliable data that makes these numbers robust. In that part of the country, from

1958 till now, there's been a 74% increase, a near doubling of the amount of precipitation that falls in

the heaviest rainfall events, the 1% of the heaviest daily events. I'm almost done, I want to show a

couple more figures.

3



2007 as I said was the big low, but then the all-time low in arctic sea-ice was last fall. This dot down

here is September of 2012, so there's a lot of year-to-year variability as you can see, but there's also a

very pronounced downward trend and if you look at the legends over here, where data began in and

beginning the satellite era in 1979, there were in in these units here 3 million square miles of arctic

sea-ice at the September minimum. That has gone down so it's now, as of last year, one-and-a-halfmillion. So we've lost half the arctic sea-ice at the September minimum, measured in terms of extent.

The volume of sea-ice is essentially the extent times the thickness of the ice, and the thickness is also

decreased so the decrease in ice volume has been greater than 50% in just this period of about 3

decades and change and the forecast now that by mid-century, perhaps 2050, 2060, there will be no

arctic sea-ice at the September minimum, which has huge implications on everything from shipping to

mineral extractions to military strategies. So we're losing arctic sea-ice very rapidly, again more rapidly

than had been projected.

3



Here are the newest projections, again, this is like a graph we saw earlier, RCP stands for these

emissions scenarios and now they've been revised so that the more serious ones, the ones that are

extending downward, have arctic sea-ice at the minimum. The black line is the same observed day that

we've seen up to 2012, but now at the minimum, you might see arctic sea-ice on the pink range of

scenarios with the highest emissions rates, RCP 8.5, it comes true, we might see arctic sea-icedisappear at the September minimum by late in the mid-century.

3



This is the last picture I want to show you, then we'll have a discussion. What you're looking at here is,

on the left, observations, and on the right, computer projections as summarized by IPCC of the

geographical distribution and the magnitude of warming over essentially the last half century. It's

actually 1957 to 2011 so this is what you have, the temperature's actually change. Remember you're

looking at the north pole region. Here is the U.S. here, here's Europe, here's Siberia. So you're lookingat the temperatures that have been observed and, using the same color code, temperatures that the

model had projected. So these models are pretty good. You can't really ignore them so you want to

attack climate science fine, but be thoughtful, don't just say that's just a computer simulation, it's not

reality because again and again and again we see that the simulations are good, and in some cases,

arctic sea-ice being an example, they've underestimated the magnitude and severity of the climate

change.

Documents

Climate Change in 4 dimensions