"We could have called this session ‘Apocalypse Now. ’ I’m going to talk to you about the consequences of climate change. Well, not ‘now’ as in right now, though, there is a part of it that’s going to wait a little longer.

So the last time I talked to you about the dynamics at work in what we usually refer to as climate change. So there’s one thing—I didn’t quite finish the overview last time—because in the substances that mankind puts in the atmosphere, there are indeed the greenhouse gases that I told you about extensively, but there is also another category of substances that men put in the atmosphere called aerosols and aerosol precursors. So an aerosol, in physical terms, is simply a suspension of liquid or solid particles in the air.

So all the people in this room who have already swept a broom, if there are any who have, you created an aerosol on that occasion, since you created a suspension of solid particles in the air, if you swept a little vigorously. So within the elements that man puts into the atmosphere which can create suspensions in the air, there actually are simultaneously elements that are emitted directly as liquids or solids, and that will then create suspensions, for example, soot. Soot is an airborne suspension of carbon particles, so it is part of aerosols.

And then there are substances that will create aerosols once in the air, that is to say, which are emitted in gaseous form and once it is in the air a reaction occurs that will transform the chemical substances into liquid or solid particles. And within this you have a very important category called sulfur dioxide, SO₂, which, once in the atmosphere, will create sulfate particles, which are bright particles particularly reflective when it comes to sunlight. And so there is this particularity, for example, the fuels with a high sulfur content, used by the merchant navy, have a cooling effect in the short term that is more important than the warming effect.

I say in the short term because aerosols are eventually purified from the atmosphere, particularly by precipitation. For example, with sulfur aerosols, it eventually results in acid rain. Sulfate dissolves in water when there is water, and it becomes something like sulfuric acid that we like a lot once it gets to the ground.

Another important effect of aerosols is that these small liquid or solid particles can act as nucleation nuclei and foster the condensation of water vapor to form clouds. And we realized, in particular, that over the areas that were highly industrialized, the clouds that formed were made up of particles that were a little thinner and precipitated a little less easily. So it has an effect, not necessarily on the cloud cover, but on the holding of this cloud cover; it has an effect.

So human emissions can also change the nature of the clouds that will end up above our heads. When the particles emitted— I come back to soot—are black, obviously it intensifies the warming because it increases the amount of solar radiation absorbed. On the other hand, when the particles are bright, typically sulfates, it is the opposite, it will increase albedo and therefore decrease the amount of solar radiation absorbed by the surface.

So these aerosols can have a warming or cooling effect, depending on their nature, but overall they all share the characteristic of being short‑lived entities. All right? These are all entities that will influence energy exchanges in the atmosphere, which are reasonably short‑term effects.

So here is an overview of the different substances that are emitted into the atmosphere, and which have an effect essentially on the reflection of solar radiation, except for soot, or black carbon. But otherwise for the most part they have a short‑term ‘cooling’ effect. This graph, which you can find in almost all editions of the IPCC reports, provides an overview of the overall warming and cooling effects of all the substances that mankind puts into the atmosphere.

So the unit used to do the radiation balance of the atmosphere is watt per square meter (W /m²). We could have used just watt, since the atmosphere has a contact surface touching the ground, which is known as the Earth’s surface. But in this case—rather than walking around with multiples of watts with 10 to the 12 power in all directions—we would rather think in terms of watt per square meter.

So if you remember the last time—well, actually even if you don’t remember well, I don’t remember if I gave you an order of magnitude—the order of magnitude of the solar radiation that reaches the ground is 200 W / m². It is a little more in fact, but it is of that order of magnitude. The order of magnitude of the natural greenhouse effect is a little less; I believe it is 170 W / m².

And the additional greenhouse effect is in the order of 2 to 3 W / m², if we measure the effects more or less in total. So here you have the additional effects, especially long‑lived greenhouse gases. Here you have a whole bunch of effects, a little more or a little less, and here you have the resultant with the error bar, which you see there, and you see that the error bar comes mainly from the aerosol part.

All right? This is mostly from there that the error bar comes from. So the overall effect of man on the exchange of energy between the atmosphere and the Earth’s surface is indeed a positive forcing effect, that is to say we are adding energy.

But you have in this story once again the aerosols that have a negative effect, but with a magnitude poorly assessed for a number of them. So once we know that we have warming and cooling effects, I will now explain the logic of a climate simulation. The logic of a climate simulation requires initially parametrizing the simulation with two inputs.

The first input is the representation of the physical system, therefore a climate model, and the second input is the hypothesis with which this physical system is disturbed, that is to say, emissions. So historically, I kept the presentation in this way because historically that is how it was done. Now that has changed a little bit; I will explain how.

Historically, the path was presented as follows: we make a hypothesis on greenhouse gas emissions, this hypothesis is itself conditioned by economic and other hypotheses—populations, etc. —and this leads to possible trajectories for future emissions. So note one important thing, it is that in the emissions scenario thus done, the emissions result from the economy that requires energy.

And so compared to what I explained to you in the first course, there is a causal inversion in the way the scenario is done, since it is usually: you have energy, and it gives you both emissions and an economy. All right? So the real entry hypothesis is not the economy that demands energy, but energy which also produces economy, that makes emissions but also makes the economy.

So we should have two outputs from the same input, which are GDP and emissions, and not emissions as a GDP output. Is it clear? Is that clear or not?

That is exactly what I explained in the first course. Nevertheless, the emission scenarios are made by economists, who therefore reverse the cause, and who tell you that there is a GDP and that GDP demands energy. In fact, to be more precise they reverse the limiting factor.

So you have emission scenarios, and these emissions go up or down, these hypotheses are assumed by principle. Once you have the emissions, you put them into a first modelling stage—so you already have a model—in which you have changes in the concentration of greenhouse gases—because actually what's forcing climate change is the evolution of the concentration of greenhouse gases. If you have increasing emissions and at the same time the sinks increase, your concentration does not move and therefore there is no forcing of the climate system.

For there to be a forcing of the climate system, there must be a thicker cover, meaning the concentration has to increase. So today in the reports that are presented by the IPCC, what you have is directly related to developments in concentration. So when you see “RCP” written somewhere now in the scientific documents that talk about climate simulations, it means “Representative Concentration Pathway”, and so these are direct hypotheses that are made about the evolution of the concentration.

And from this can be drawn possible emission scenarios. This evolution of concentration is therefore given as an input to a climate model, and this climate model will say ‘ouch’ more or less louder, and it will tell you how it makes ‘ouch. ’ All right?

We are going to spend a lot of time today watching this. So the climate model will tell you “it’s getting hot,” “it’s burning, here,” “it’s getting cold, there,” “it hurts,” and “here I’m getting a flood,” “here a hurricane,” “the hurricane is more complicated,” etc. So the climate model will tell you what it sees in the evolution of the climate system.

In the most advanced models, now there is feedback between the concentration and the climate model. So an example: if the climate drifts too quickly and ecosystems that cannot keep up, some of the CO₂ sinks weaken, ecosystems die, so there is less photosynthesis, CO₂ is less well recovered from the atmosphere, and therefore with identical emissions, the concentration increases more quickly. These closed‑loop phenomena are taken into account in the most recent models.

However, what even the most recent models do not take into account, and in my opinion will never take into account—even those who claim to include coupling with the economy—is feedback on input hypotheses, that is, there is not a model that tells you: “Here, the climate system has drifted so fast, that it is not compatible with maintaining the current GDP. ” There is no climate model that tells that story. So with this particularity, or this oddity, the simulations that give you the most unpleasant results in terms of global warming go with the most important emission scenarios that, in the way they are written today, also go with the economy that is doing best.

So we have something a little surprising: the more the climate system is disrupted, the better the economy is doing. That is what you have, chronologically, in the way simulations are done today. Do I repeat or is it clear?

The highest level of emission is needed, that is to say the healthiest economy, in the stories that are being told, to make the climate more unstable. Which is another way of saying that the climate will never, in the way simulations are done, harm the economic system, by construction. So how future emissions are going to be is obviously impossible to predict.

I personally don’t know how many men there will be in 40 years, I don’t know how much oil per person, and gas, and coal per person, and I don’t know how much meat people will eat per person. All I can say is that if it is going to be different from what it is today, that means there will already have been a disaster that will have been responsible for quickly regulating the system. That’s the only thing I know I can say.

I cannot predict anything else. So people who run simulations don’t know how to predict, so they run scenarios. So they take several possibilities, and in what I will show you from now on, the results will always be subject to the hypothesis we chose at the beginning.

This is something essential. It must be understood that the results you will see from now on are always subject to emissions conditions, and therefore—since I think I will not repeat this subject for this year’s examination—one of the reasons why it is impossible to accurately predict the risks of climate change is that it is impossible to accurately predict what the emissions will be. I asked your friends last year: “Why is it impossible to accurately predict the risks of climate change?

” I think only one out of five of them— in the copies I've already managed to correct—told me that it was because the emissions were not predictable. That’s it—either the other four were asleep, or that day they were out of shape. So since we cannot predict emissions, we will never be able to accurately predict the consequences.

Here is a perspective of the first stage of the modelling shows, i. e. moving from emissions to concentration.

Once again, I would remind you that we are going backwards. First we have “RCP”, and from there we derive the possible trajectories on emissions. That is how it is done today.

You have here, therefore, a perspective on the concentration— in this case the most important of them— of CO₂ in the air. The one that began to warp until the end of the 20th century. And here you have the possible evolutions during the 21st century, but you see that these possible evolutions are very different according to the emission scenarios.

So the low scenario, so far, I take it as emissions that remain constant. So actually, I should present you the 1. 5 scenario, for example.

Basically, the 1. 5 scenario, is about what we already emitted, the emissions do not remain constant, the emissions turn to zero tomorrow morning. That is the 1.

5 scenario. From emissions that remain constant until emissions that are very significantly increased. So until now I used to call it my “Polish of the Year 2000” scenario.

Why “Polish of the Year 2000”? Because the Polish is the European country which still relies most heavily on coal today, in proportion. And on the other hand, “of the Year 2000”, because this scenario corresponds to the consumption of 10 billion people who would use, per person, the same amount of energy as a Polish today.

That is why I called it that, and it gives you a bit of a socio‑economic picture of what such an emissions scenario means. So in fact, it just means a humanity of 10 billion people who consider that what is acceptable, what is normal, what is standard, is to live like a Polish of the year 2000. If we have that, both in terms of energy consumption and technical performance, we come up with an emissions scenario that gives you this CO₂ concentration at the end of the 21st century, to give you an idea.

Once we have an evolution of the CO₂ concentration, we put it in a climate model. So, what is a climate model? It is a virtual planet— instead of being a flight simulator, it is a climate simulator— and this virtual planet, which is a digital model that works on a computer, is necessarily a discrete model because you know that even very powerful computers do not like infinity.

All they know is to manage a finite number of operations or a finite number of points. Yes? ***Audience question*** Is it the worst scenario you can make?

For instance, isn’t a Chinese of today worse than a Polish? ***Audience question*** We can always make worse scenarios, there is no problem. I am just saying that this one corresponds to a Polish from the year 2000 applied to ten billion people.

***Audience intervention*** Then, why would you choose this scenario? ***Audience intervention*** I chose this. It's just to illustrate.

Because you still have a standard of living that is higher than a Polish today, I say that if 10 billion people on Earth want to live a little worse than you, that is the emission scenario. Or with the level of material comfort you have today to be more precise, that is what it gives in terms of emissions. And so as a level of concentration.

It was to illustrate that. So I was saying, a climate model is a numerical representation of the planet, it is a model, and therefore by construction, it is something that is discrete. So there are meshes.

This is how it works: all the fluid compartments on the planet are divided into shoe boxes, i. e. parallelepipeds, or almost, because they are a little curved.

They are parallelepipeds for clown or for people with very flat feet, because typically the length on the horizontal is a few tens to 100 km, we will say 100 km in order of magnitude, on the other hand the thickness on the vertical is rather on the order of a hundred meters. So they are very, very flat feet. But otherwise, that is the idea: we are going to cut the atmosphere and the ocean into shoe boxes, and we are going to represent the soil with meshes, with properties of exchange with the atmosphere that are different depending on whether it is water or land, obviously.

These models have been sophisticated over time but all are based on the basic equations of physics. And then an important thing—it may seem very surprising at first because you say to yourself, “Well, that's not how you usually experiment”—but, one thing that is very important to know is that what you look at with these models is how they converge with given greenhouse gas concentrations, and it does not depend on the initial state with which they are configured. In other words, with these models, whether the average temperature, at year zero, is 15, 16 or 17 degrees, does not matter.

In fact, what we are going to look at is how they converge over time, with what we have given them as an evolution on the amount of greenhouse gases included in the atmosphere. So these models have improved over time. The first thing they did was that the mesh—because actually, they get better as the available computing power increases.

You know that traditionally, weather agencies are among the buyers of supercomputers. The people who like very powerful computers are the people from the weather channel, and when it comes to climate models, they run with an atmospheric part that is exactly the same as the weather channel. Then they are less fine meshes but they are the same equations.

The four acronyms or abbreviations you have here are representative of the first four IPCC assessment reports. F for “First Assessment Report”, S for “Second”, T for “Third”, and as F had already been used for “First”, it was not reuse F for “Fourth”, so “AR4” was chosen. We thought that sometimes people get confused, it's not completely impossible.

You see two things. First, the mesh size has decreased very significantly, and the second thing you see is that some relief has appeared. That is, initially the ground was flat, and in fact, what was different was the water and the land.

And on land, you had vegetation and uncovered land that was differentiated, but it was not differentiated in the carbon cycle, it was mainly differentiated in the albedo, that is, the reflectivity of the surface. And then you see that with time, relief starts to appear, that is to say that we show an ability to differentiate in altitude what occupies a mesh, whether it is the atmosphere or the ground. Yes?

***Audience question*** No, it is related to the fact that, on the first one, you see that it is flat. Yes, there is a little thing on the right, if you want. Oh yes, indeed, on the left you have topography, excuse me.

But it is a topography that remains very rough. You are right. The second thing about the models is that they progressively became more complex in terms of the processes that are taken into account.

At the time of the first generation of models—let’s say— they were essentially atmospheric, and the rest of the compartments of the planet did not appear and therefore the exchanges with the ocean were set up, that is, there was no ocean dynamics. The ocean was seen as a surface of exchange with the planet, and exchanges were set to be constant. And then as the generations of models progress, more and more variables appear.

So ocean, vegetation, the ocean dynamic, aerosols. And now, in the most recent models, there is most of what happens in the atmosphere— on Earth, to be exact— in the atmosphere and in the ocean. You obviously see the cloud cover that has appeared.

Now there is atmospheric chemistry, you have the carbon cycle, etc. You have most of the major processes that are represented. Something interesting to note, by the way, is that the improvement of these models, never has fundamentally changed the orders of magnitude on the temperature rise resulting from a given increase in CO₂ concentration in the atmosphere.

Quite simply because this temperature rise results above all from the amount of additional energy that is sent to the ground by an additional greenhouse effect. And it is a fairly well understood physical process of absorption‑release by a gas. All right?

Once you have the total amount of energy returning to the ground, and you know how to differentiate what is absorbed by the ocean and what is absorbed by the ground, right away, you have your ground temperature rise quite easily. So the sophistication of the models makes possible to have a better view for a certain number of local phenomena, in particular. It allows to improve the spatial resolution of what will happen.

However, it has not changed in the first place— I insist— the overall consequences of what is going to happen, and in particular the rise in average temperature, which represents the increase in the amount of energy contained in the system. Climate models—like all models when they are used by people who know pretty much what they are doing— the first thing you do once you have developed them is to test them on the past. When you have a model, the first thing you do to make sure it works is to confront it with what happened before.

By the way, there is not an economic model that passes the test of this kind of thing, unless you change the parameters year after year to make sure it fits, but otherwise it does not work. As far as physical models are concerned, that is exactly how they are calibrated, and that is exactly how climate models are calibrated. What you see here— as in any IPCC graph, there is a lot of information— is how a set of models— from memory there are about twenty of them in there— compares with what has been observed with respect to temperature changes.

So, with regard to the evolution of temperatures, on each graph, which represent either an area here or the whole world of land or oceans below, you have what the models tell you when they are allowed to evolve with an evolution of greenhouse gases that has been what it is— in pink. In black, it is what the temperature readings really showed, everywhere. And so what you see is that with the exception of a few anomalies, at that time, for example on the ocean— and this anomaly repeats itself a little on every single graph there— models reproduce quite well the general trend on temperature variation over the period.

So, the same modelers had fun looking at what the model would have given— which reproduces what happened— if greenhouse gas emissions had not increased. And in this case, it gives you the blue curve, or more precisely the blue envelope. So the blue envelope is the envelope of all simulations that are done with climate models in a virtual world where humans did not emit greenhouse gases.

And in that world, you can see that overall temperatures remain more or less constant. The system is almost stable, whereas when you add the greenhouse gas emissions, the system drifts away. This kind of test is absolutely essential to give credibility to these digital tools for what they will tell later.

I did not bring it to you today, but you can find simulations on the Internet— for those of you who will be looking for it— in this case, in real time, of these climate models that simulate the Earth seen from the atmosphere. There is a comparison between real satellite images on the one hand, and a representation with a digital model on the other. And when you compare the two— test it with someone in double blind— in general, people cannot say which one is which.

Today, there is also an extremely fine quality of representation with regard to the ability of these models to visually represent the system. So we will talk about the future with these models, once they have been able to trace the past. What you have here is again a graph that comes from the last big IPCC report— the one before of 2007, sorry, not 2013, but it has not changed much in the last one.

The advantage of this graph is that it discriminates very precisely how the temperature can change depending on the emission scenario. That’s the idea you have to have in mind. So you can see here the evolution of temperature over the 20th century.

You can see that the temperature evolution over the 20th century has already increased by a fraction of a degree. And then, the way in which the temperature changes in the future according to the hypothesis made about greenhouse gas emissions. So, the lowest assumption in these simulations is that the concentration remains constant.

This means that, basically, emissions start to decrease very, very quickly starting tomorrow morning, and they reach zero well before the end of the century. So now, indeed, you see that we are holding the 1. 5°C.

So here, you have one degree. We started about there, we are holding the 1. 5°C.

In other words, to hold the 1. 5°C, emissions must start to fall very quickly starting tomorrow, and reach zero well before the end of the century. That’s what it means to hold the 1.

5°C. And that is not what we are heading towards. Then, there is this scenario here in which we keep emissions basically constant.

So you see that this is a scenario in which we gain 2°C compared to 1990, which you have here, but we have already gained almost 3°C compared to the pre‑industrial era. So what you have here is a scenario that is almost 3°C. What you see here is the envelope for this scenario for all simulations, all models.

In other words, of the twenty models used with this same scenario, the most conservative one tells you: “we will gain one degree compared to 1990”, and the most pessimistic one tells you: “we will gain 3 degrees compared to 1990. ” This graph tells you that there is still variability related to the assumptions we made in the model because between all the models there are not exactly the same options that are taken for the representation of a certain number of processes, or more exactly for the values taken for the parameters, or the boundary conditions of a certain number of processes, because there is stuff that remains set in there. And of course, we try to adjust the parameters according to past observations and thus we try to recalibrate the model continuously.

So depending on the model that is used, you can see that you still have a margin of error of about 2°C on this scenario on the temperature increase you have at the end. So we are not quite in the thickness of the line because you will see that the consequences related to climate change are not at all the same at +1°C and +3°C. The middle scenario here is a scenario that is considered a coal scenario— at least in my jargon— that is, emissions double.

And then, there, always the same, which corresponds to this scenario is there, and so in the low version we are only at 1. 5°C but in the high version we are at more than 4°C. So very often there is a mistake made in the face of this kind of uncertainty, which is to say “if there is still a great uncertainty about the fact that we are in great danger, then I will wait to know more to be sure that it is really a great danger”.

So the parallel I am going to take is if one of your friends, one day when he's had a lot to drink, asks you to cross the Champs Élysées at 5 a. m. , blindfolded, I think there's a big uncertainty as to whether you are going to die or not: I’m not completely sure you are going to do that.

So actually, a great uncertainty is a great risk. If there is a great uncertainty, it means that I am not able to give you an upper bound on what you are risking. All right?

So the idea that a great uncertainty is a great insurance policy that means that it is urgent to do nothing is the exact opposite of the conclusion to be drawn. A great uncertainty is a great risk, a very great risk. And finally, my Polish scenario is a scenario that you can see can take us beyond 6°C with the generation of models that were used at that time.

So now, we usually say it is more like 5°C, but in the end, it does not change much the conclusions and the implications. So you understand, here, that you have a dispersion once again linked to the model used for the same scenario. Another way of presenting the same thing is that when you run a simulation— you can do two things in fact.

Yes? ***Audience question*** The 2°C objective, is it from the year 2000 or 1900? No, it is from 1850.

And so, there is already one".